zdoom/gzdoom-gles
0
0
Fork 0
mirror of https://github.com/ZDoom/gzdoom-gles.git synced 2025-01-09 10:40:56 +00:00
Code Issues Projects Releases Packages Wiki Activity
gzdoom-gles/src/scripting/zscript/zcc_compile.cpp

3281 lines
94 KiB
C++
Raw Normal View History

- added copyright headers to scripting files.
2016-10-08 06:35:16 +00:00
/*
** zcc_compile.cpp
**
**---------------------------------------------------------------------------
** Copyright -2016 Randy Heit
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
** Copyright 2016 Christoph Oelckers
- added copyright headers to scripting files.
2016-10-08 06:35:16 +00:00
** All rights reserved.
**
** Redistribution and use in source and binary forms, with or without
** modification, are permitted provided that the following conditions
** are met:
**
** 1. Redistributions of source code must retain the above copyright
** notice, this list of conditions and the following disclaimer.
** 2. Redistributions in binary form must reproduce the above copyright
** notice, this list of conditions and the following disclaimer in the
** documentation and/or other materials provided with the distribution.
** 3. The name of the author may not be used to endorse or promote products
** derived from this software without specific prior written permission.
**
** THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
** IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
** OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
** IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
** INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
** NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
** DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
** THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
** (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
** THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**---------------------------------------------------------------------------
**
*/
- reviewed all places where VARF_Action and NAP were used and fixed what was still wrong.
2016-10-22 16:13:54 +00:00
#include "a_pickups.h"
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
#include "thingdef.h"
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
#include "sc_man.h"
#include "c_console.h"
#include "c_dispatch.h"
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
#include "doomerrors.h"
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
#include "w_wad.h"
#include "cmdlib.h"
#include "m_alloc.h"
#include "zcc_parser.h"
- process most of the remaining statement types. - extend assignment operations to handle local variables (untested.) - make the implicit function arguments read only.
2016-10-20 23:12:54 +00:00
#include "zcc-parse.h"
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
#include "zcc_compile.h"
#include "v_text.h"
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
#include "p_lnspec.h"
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
#include "i_system.h"
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
#include "gdtoa.h"
- committed from the wrong subdirectory and missed all files that just changed the #include line.
2016-10-15 08:43:02 +00:00
#include "codegeneration/codegen.h"
- separated the code generation from the DECORATE parser and cleaned up the interface to the code generator. Most importantly, the VMScriptFunctions are now preallocated when being added to the list of functions to compile and will be filled in later by the code generator. This allowed the removal of some ugly maintenance code.
2016-10-12 22:53:59 +00:00
#include "vmbuilder.h"
- zcc_compile.cpp needs to include version.h for the DYNLIGHT #define.
2016-12-03 18:14:20 +00:00
#include "version.h"
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
//==========================================================================
//
// ZCCCompiler :: ProcessClass
//
//==========================================================================
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
void ZCCCompiler::ProcessClass(ZCC_Class *cnode, PSymbolTreeNode *treenode)
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
{
- created an export for P_SpawnMissile so that I can do some tests with functions. - allow class extensions. These are separate blocks in different files that get concatenated to one class body for processing. The reason is to allow spreading the many functions in Actor over multiple files, so that they remain manageable. For example, all the Doom action functions should be in their respective files, but their symbols need to be in Actor. To extend a class, both files need to be in the same translation unit, so it won't allow user-side extension of internal classes.
2016-10-23 10:57:21 +00:00
ZCC_ClassWork *cls = nullptr;
// If this is a class extension, put the new node directly into the existing class.
if (cnode->Flags == ZCC_Extension)
{
for (auto clss : Classes)
{
if (clss->NodeName() == cnode->NodeName)
{
cls = clss;
break;
}
}
if (cls == nullptr)
{
Error(cnode, "Class %s cannot be found in the current translation unit.");
return;
}
}
else
{
Classes.Push(new ZCC_ClassWork(static_cast<ZCC_Class *>(cnode), treenode));
cls = Classes.Last();
}
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
auto node = cnode->Body;
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
PSymbolTreeNode *childnode;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
ZCC_Enum *enumType = nullptr;
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// Need to check if the class actually has a body.
if (node != nullptr) do
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
{
switch (node->NodeType)
{
case AST_Struct:
case AST_ConstantDef:
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if ((childnode = AddTreeNode(static_cast<ZCC_NamedNode *>(node)->NodeName, node, &cls->TreeNodes)))
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
{
switch (node->NodeType)
{
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
enumType = static_cast<ZCC_Enum *>(node);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Enums.Push(enumType);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
case AST_Struct:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
ProcessStruct(static_cast<ZCC_Struct *>(node), childnode, cls->cls);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
case AST_ConstantDef:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Constants.Push(static_cast<ZCC_ConstantDef *>(node));
cls->Constants.Last()->Type = enumType;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
default:
assert(0 && "Default case is just here to make GCC happy. It should never be reached");
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
}
}
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_VarDeclarator:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Fields.Push(static_cast<ZCC_VarDeclarator *>(node));
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
case AST_EnumTerminator:
enumType = nullptr;
break;
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
case AST_States:
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
cls->States.Push(static_cast<ZCC_States *>(node));
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
break;
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
case AST_FuncDeclarator:
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
cls->Functions.Push(static_cast<ZCC_FuncDeclarator *>(node));
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
break;
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
case AST_Default:
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
cls->Defaults.Push(static_cast<ZCC_Default *>(node));
break;
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
default:
assert(0 && "Unhandled AST node type");
break;
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
}
node = node->SiblingNext;
}
while (node != cnode->Body);
}
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
//==========================================================================
//
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// ZCCCompiler :: ProcessStruct
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
//
//==========================================================================
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
void ZCCCompiler::ProcessStruct(ZCC_Struct *cnode, PSymbolTreeNode *treenode, ZCC_Class *outer)
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Structs.Push(new ZCC_StructWork(static_cast<ZCC_Struct *>(cnode), treenode, outer));
ZCC_StructWork *cls = Structs.Last();
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
auto node = cnode->Body;
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
PSymbolTreeNode *childnode;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
ZCC_Enum *enumType = nullptr;
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// Need to check if the struct actually has a body.
if (node != nullptr) do
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
{
switch (node->NodeType)
{
case AST_ConstantDef:
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if ((childnode = AddTreeNode(static_cast<ZCC_NamedNode *>(node)->NodeName, node, &cls->TreeNodes)))
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
{
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
switch (node->NodeType)
{
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
enumType = static_cast<ZCC_Enum *>(node);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Enums.Push(enumType);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
case AST_ConstantDef:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Constants.Push(static_cast<ZCC_ConstantDef *>(node));
cls->Constants.Last()->Type = enumType;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
default:
assert(0 && "Default case is just here to make GCC happy. It should never be reached");
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
}
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
}
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_VarDeclarator:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
cls->Fields.Push(static_cast<ZCC_VarDeclarator *>(node));
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
case AST_FuncDeclarator:
cls->Functions.Push(static_cast<ZCC_FuncDeclarator *>(node));
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_EnumTerminator:
enumType = nullptr;
break;
- added ProcessStruct, analogous to ProcessClass. Now everything should be in place to resolve all constants - inside and outside classes or structs.
2016-10-08 13:39:54 +00:00
default:
assert(0 && "Unhandled AST node type");
break;
}
node = node->SiblingNext;
}
while (node != cnode->Body);
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//==========================================================================
//
// ZCCCompiler Constructor
//
//==========================================================================
- added a new variable flag that allows defining variables which can only be modified by internal script code but not by external mods. Currently this is used by the NoBlockmap and NoSector flags which need special handling for changing.
2016-11-08 10:12:56 +00:00
ZCCCompiler::ZCCCompiler(ZCC_AST &ast, DObject *_outer, PSymbolTable &_symbols, PSymbolTable &_outsymbols, int lumpnum)
: Outer(_outer), GlobalTreeNodes(&_symbols), OutputSymbols(&_outsymbols), AST(ast), Lump(lumpnum)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- count errors in the parser and abort afterward if there were some. - use FScriptPosition's error counter throughout the compiler so that there is only one counter for everything, not two. Parts of the compiler use FScriptPosition, so this is easier to handle than having a separate counter in the compiler class. It also avoids having to pass the compiler object to any function where an error may be output. The TreeNodes contain sufficient data to be converted to an FScriptPosition and using that for error message formatting.
2016-10-15 10:15:25 +00:00
FScriptPosition::ResetErrorCounter();
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
// Group top-level nodes by type
if (ast.TopNode != NULL)
{
ZCC_TreeNode *node = ast.TopNode;
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
PSymbolTreeNode *tnode;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
PType *enumType = nullptr;
ZCC_Enum *zenumType = nullptr;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
do
{
switch (node->NodeType)
{
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
case AST_Class:
- created an export for P_SpawnMissile so that I can do some tests with functions. - allow class extensions. These are separate blocks in different files that get concatenated to one class body for processing. The reason is to allow spreading the many functions in Actor over multiple files, so that they remain manageable. For example, all the Doom action functions should be in their respective files, but their symbols need to be in Actor. To extend a class, both files need to be in the same translation unit, so it won't allow user-side extension of internal classes.
2016-10-23 10:57:21 +00:00
// a class extension should not check the tree node symbols.
if (static_cast<ZCC_Class *>(node)->Flags == ZCC_Extension)
{
ProcessClass(static_cast<ZCC_Class *>(node), tnode);
break;
}
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
case AST_Struct:
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
case AST_ConstantDef:
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if ((tnode = AddTreeNode(static_cast<ZCC_NamedNode *>(node)->NodeName, node, GlobalTreeNodes)))
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
switch (node->NodeType)
{
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_Enum:
zenumType = static_cast<ZCC_Enum *>(node);
enumType = NewEnum(zenumType->NodeName, nullptr);
GlobalSymbols.AddSymbol(new PSymbolType(zenumType->NodeName, enumType));
break;
case AST_Class:
ProcessClass(static_cast<ZCC_Class *>(node), tnode);
break;
case AST_Struct:
ProcessStruct(static_cast<ZCC_Struct *>(node), tnode, nullptr);
break;
case AST_ConstantDef:
Constants.Push(static_cast<ZCC_ConstantDef *>(node));
Constants.Last()->Type = zenumType;
break;
default:
assert(0 && "Default case is just here to make GCC happy. It should never be reached");
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
}
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case AST_EnumTerminator:
zenumType = nullptr;
break;
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
default:
assert(0 && "Unhandled AST node type");
break;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
node = node->SiblingNext;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
} while (node != ast.TopNode);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
}
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
ZCCCompiler::~ZCCCompiler()
{
for (auto s : Structs)
{
delete s;
}
for (auto c : Classes)
{
delete c;
}
Structs.Clear();
Classes.Clear();
}
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
//==========================================================================
//
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
// ZCCCompiler :: AddTreeNode
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
//
// Keeps track of definition nodes by their names. Ensures that all names
// in this scope are unique.
//
//==========================================================================
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
PSymbolTreeNode *ZCCCompiler::AddTreeNode(FName name, ZCC_TreeNode *node, PSymbolTable *treenodes, bool searchparents)
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
PSymbol *check = treenodes->FindSymbol(name, searchparents);
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
if (check != NULL)
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
assert(check->IsA(RUNTIME_CLASS(PSymbolTreeNode)));
Error(node, "Attempt to redefine '%s'", name.GetChars());
Error(static_cast<PSymbolTreeNode *>(check)->Node, " Original definition is here");
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
return nullptr;
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
}
else
{
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
auto sy = new PSymbolTreeNode(name, node);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
FString name;
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
treenodes->AddSymbol(sy);
- give the 'default' block its own node type in the AST. - group a class's body the same way as the global AST. Only partially implemented so far so that constants can be processed.
2016-10-08 13:27:20 +00:00
return sy;
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
}
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//==========================================================================
//
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
// ZCCCompiler :: Warn
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
// Prints a warning message, and increments WarnCount.
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
//==========================================================================
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
void ZCCCompiler::Warn(ZCC_TreeNode *node, const char *msg, ...)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
va_list argptr;
va_start(argptr, msg);
MessageV(node, TEXTCOLOR_ORANGE, msg, argptr);
va_end(argptr);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
- count errors in the parser and abort afterward if there were some. - use FScriptPosition's error counter throughout the compiler so that there is only one counter for everything, not two. Parts of the compiler use FScriptPosition, so this is easier to handle than having a separate counter in the compiler class. It also avoids having to pass the compiler object to any function where an error may be output. The TreeNodes contain sufficient data to be converted to an FScriptPosition and using that for error message formatting.
2016-10-15 10:15:25 +00:00
FScriptPosition::WarnCounter++;
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
//==========================================================================
//
// ZCCCompiler :: Error
//
// Prints an error message, and increments ErrorCount.
//
//==========================================================================
void ZCCCompiler::Error(ZCC_TreeNode *node, const char *msg, ...)
{
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
va_list argptr;
va_start(argptr, msg);
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
MessageV(node, TEXTCOLOR_RED, msg, argptr);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
va_end(argptr);
- count errors in the parser and abort afterward if there were some. - use FScriptPosition's error counter throughout the compiler so that there is only one counter for everything, not two. Parts of the compiler use FScriptPosition, so this is easier to handle than having a separate counter in the compiler class. It also avoids having to pass the compiler object to any function where an error may be output. The TreeNodes contain sufficient data to be converted to an FScriptPosition and using that for error message formatting.
2016-10-15 10:15:25 +00:00
FScriptPosition::ErrorCounter++;
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
//==========================================================================
//
// ZCCCompiler :: MessageV
//
// Prints a message, annotated with the source location for the tree node.
//
//==========================================================================
void ZCCCompiler::MessageV(ZCC_TreeNode *node, const char *txtcolor, const char *msg, va_list argptr)
{
FString composed;
composed.Format("%s%s, line %d: ", txtcolor, node->SourceName->GetChars(), node->SourceLoc);
composed.VAppendFormat(msg, argptr);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
composed += '\n';
PrintString(PRINT_HIGH, composed);
}
//==========================================================================
//
// ZCCCompiler :: Compile
//
// Compile everything defined at this level.
//
//==========================================================================
int ZCCCompiler::Compile()
{
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
CreateClassTypes();
CreateStructTypes();
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
CompileAllConstants();
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
CompileAllFields();
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
InitDefaults();
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
InitFunctions();
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
CompileStates();
- count errors in the parser and abort afterward if there were some. - use FScriptPosition's error counter throughout the compiler so that there is only one counter for everything, not two. Parts of the compiler use FScriptPosition, so this is easier to handle than having a separate counter in the compiler class. It also avoids having to pass the compiler object to any function where an error may be output. The TreeNodes contain sufficient data to be converted to an FScriptPosition and using that for error message formatting.
2016-10-15 10:15:25 +00:00
return FScriptPosition::ErrorCounter;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
//==========================================================================
//
// ZCCCompiler :: CreateStructTypes
//
// Creates a PStruct for every struct.
//
//==========================================================================
void ZCCCompiler::CreateStructTypes()
{
for(auto s : Structs)
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
s->Outer = s->OuterDef == nullptr? nullptr : s->OuterDef->CType();
- implemented State as an actual native struct, so that its fields can be accessed from scripts. - refactored state bitfield members into a flag word because the address of a bitfield cannot be taken, making such variables inaccessible to scripts. - actually use PNativeStruct for representing native structs defined in a script.
2016-11-21 20:34:34 +00:00
if (s->strct->Flags & ZCC_Native)
s->strct->Type = NewNativeStruct(s->NodeName(), nullptr);
else
s->strct->Type = NewStruct(s->NodeName(), s->Outer);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
s->strct->Symbol = new PSymbolType(s->NodeName(), s->Type());
GlobalSymbols.AddSymbol(s->strct->Symbol);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
for (auto e : s->Enums)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto etype = NewEnum(e->NodeName, s->Type());
s->Type()->Symbols.AddSymbol(new PSymbolType(e->NodeName, etype));
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
}
}
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
//==========================================================================
//
// ZCCCompiler :: CreateClassTypes
//
// Creates a PClass for every class so that we get access to the symbol table
// These will be created with unknown size because for that we need to
// process all fields first, but to do that we need the PClass and some
// other info depending on the PClass.
//
//==========================================================================
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
void ZCCCompiler::CreateClassTypes()
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
// we are going to sort the classes array so that entries are sorted in order of inheritance.
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
auto OrigClasses = std::move(Classes);
Classes.Clear();
bool donesomething = true;
while (donesomething)
{
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
donesomething = false;
for (unsigned i = 0; i<OrigClasses.Size(); i++)
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
{
auto c = OrigClasses[i];
// Check if we got the parent already defined.
PClass *parent;
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto ParentName = c->cls->ParentName;
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (ParentName != nullptr && ParentName->SiblingNext == ParentName) parent = PClass::FindClass(ParentName->Id);
else if (ParentName == nullptr) parent = RUNTIME_CLASS(DObject);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
else
{
// The parent is a dotted name which the type system currently does not handle.
// Once it does this needs to be implemented here.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto p = ParentName;
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
FString build;
do
{
if (build.IsNotEmpty()) build += '.';
build += FName(p->Id);
p = static_cast<decltype(p)>(p->SiblingNext);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
} while (p != ParentName);
Error(c->cls, "Qualified name '%s' for base class not supported in '%s'", build.GetChars(), FName(c->NodeName()).GetChars());
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
parent = RUNTIME_CLASS(DObject);
}
if (parent != nullptr)
{
// The parent exists, we may create a type for this class
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (c->cls->Flags & ZCC_Native)
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
{
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
// If this is a native class, its own type must also already exist and not be a runtime class.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto me = PClass::FindClass(c->NodeName());
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
if (me == nullptr)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(c->cls, "Unknown native class %s", c->NodeName().GetChars());
- fixed: ZScript did not properly fill in tentatively created classes from defaults processing. The proper setup for such classes was only done in CreateDerivedClass, but not in FindClassTentative itself. This extends CreateDerivedClass to allow it to create a class without fully initializing it.
2016-11-12 12:09:19 +00:00
// Create a placeholder so that the compiler can continue looking for errors.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
me = parent->FindClassTentative(c->NodeName());
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
}
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
else if (me->bRuntimeClass)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(c->cls, "%s is not a native class", c->NodeName().GetChars());
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
}
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
else
{
DPrintf(DMSG_SPAMMY, "Registered %s as native with parent %s\n", me->TypeName.GetChars(), parent->TypeName.GetChars());
}
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
c->cls->Type = me;
- fixed: String comparisons with constants did not work.
2016-11-09 19:28:46 +00:00
auto ac = dyn_cast<PClassActor>(me);
if (ac != nullptr) ac->SourceLumpName = *c->cls->SourceName;
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
}
else
{
- create struct types. - fixed class creation. There was an infinite loop and some missing checks for native classes. - do not write the compiler's symbols to the same symbol table as the output. The output must go to GlobalSymbols but the internal symbols must go to a namespace specific table that can be discarded after compilation.
2016-10-08 11:34:37 +00:00
// We will never get here if the name is a duplicate, so we can just do the assignment.
- fixed: ZScript did not properly fill in tentatively created classes from defaults processing. The proper setup for such classes was only done in CreateDerivedClass, but not in FindClassTentative itself. This extends CreateDerivedClass to allow it to create a class without fully initializing it.
2016-11-12 12:09:19 +00:00
try
{
c->cls->Type = parent->CreateDerivedClass(c->NodeName(), TentativeClass);
if (c->Type() == nullptr)
{
Error(c->cls, "Class name %s already exists", c->NodeName().GetChars());
}
}
catch (CRecoverableError &err)
{
Error(c->cls, "%s", err.GetMessage());
// create a placeholder so that the compiler can continue looking for errors.
c->cls->Type = nullptr;
}
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
}
- fixed: ZScript did not properly fill in tentatively created classes from defaults processing. The proper setup for such classes was only done in CreateDerivedClass, but not in FindClassTentative itself. This extends CreateDerivedClass to allow it to create a class without fully initializing it.
2016-11-12 12:09:19 +00:00
if (c->Type() == nullptr) c->cls->Type = parent->FindClassTentative(c->NodeName());
- scriptified parts of a_bossbrain.cpp. Some things cannot be done yet, the script code is there but commented out. - exported thinker iterator and drop item chain to scripting. Unlike its native counterpart the script-side iterator is wrapped into a DObject to allow proper handling for memory management. - fixed: The VMFunctionBuilder only distinguished between member and action functions but failed on static ones. - fixed: FxAssign did not add all needed type casts. Except for purely numeric types it will now wrap the expression in an FxTypeCast. Numeric handling remains unchanged for both performance reasons and not altering semantics for DECORATE. - exported all internal flags as variables to scripting. They still cannot be used in an actor definition. - make ATAG_STATE the same as ATAG_GENERIC. Since state pointers exist as actual variables they can take both values which on occasion can trigger some asserts. - gave PClass a bExported flag, so that scripts cannot see purely internal classes. Especially the types like PInt can cause problems. Todo: we need readonly references to safely expose the actor defaults. Right now some badly behaving code could overwrite them.
2016-10-31 16:02:47 +00:00
c->Type()->bExported = true; // this class is accessible to script side type casts. (The reason for this flag is that types like PInt need to be skipped.)
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
c->cls->Symbol = new PSymbolType(c->NodeName(), c->Type());
GlobalSymbols.AddSymbol(c->cls->Symbol);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
Classes.Push(c);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
OrigClasses.Delete(i--);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
donesomething = true;
}
else
{
// No base class found. Now check if something in the unprocessed classes matches.
// If not, print an error. If something is found let's retry again in the next iteration.
bool found = false;
for (auto d : OrigClasses)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (d->NodeName() == c->cls->ParentName->Id)
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
{
found = true;
break;
}
}
if (!found)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(c->cls, "Class %s has unknown base class %s", c->NodeName().GetChars(), FName(c->cls->ParentName->Id).GetChars());
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
// create a placeholder so that the compiler can continue looking for errors.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
c->cls->Type = RUNTIME_CLASS(DObject)->FindClassTentative(c->NodeName());
c->cls->Symbol = new PSymbolType(c->NodeName(), c->Type());
GlobalSymbols.AddSymbol(c->cls->Symbol);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
Classes.Push(c);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
OrigClasses.Delete(i--);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
donesomething = true;
}
}
}
}
// What's left refers to some other class in the list but could not be resolved.
// This normally means a circular reference.
for (auto c : OrigClasses)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(c->cls, "Class %s has circular inheritance", FName(c->NodeName()).GetChars());
c->cls->Type = RUNTIME_CLASS(DObject)->FindClassTentative(c->NodeName());
c->cls->Symbol = new PSymbolType(c->NodeName(), c->Type());
GlobalSymbols.AddSymbol(c->cls->Symbol);
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
Classes.Push(c);
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
- fixed: ZCCCompiler tried to process replacements too early, it can only be done after all classes have been created.
2016-11-04 14:16:56 +00:00
// Last but not least: Now that all classes have been created, we can create the symbols for the internal enums and link the treenode symbol tables and set up replacements
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
for (auto cd : Classes)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
for (auto e : cd->Enums)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto etype = NewEnum(e->NodeName, cd->Type());
cd->Type()->Symbols.AddSymbol(new PSymbolType(e->NodeName, etype));
}
// Link the tree node tables. We only can do this after we know the class relations.
for (auto cc : Classes)
{
if (cc->Type() == cd->Type()->ParentClass)
{
cd->TreeNodes.SetParentTable(&cc->TreeNodes);
break;
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- fixed: ZCCCompiler tried to process replacements too early, it can only be done after all classes have been created.
2016-11-04 14:16:56 +00:00
if (cd->cls->Replaces != nullptr && !static_cast<PClassActor *>(cd->Type())->SetReplacement(cd->cls->Replaces->Id))
{
Warn(cd->cls, "Replaced type '%s' not found for %s", FName(cd->cls->Replaces->Id).GetChars(), cd->Type()->TypeName.GetChars());
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- create class types from the AST. This is the first thing the compiler has to do to get access to the class's symbol table. Of course at this point the final size of a class is not known yet so these are currently all treated as tentative.
2016-10-08 09:54:33 +00:00
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//==========================================================================
//
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
// ZCCCompiler :: AddConstants
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
// Helper for CompileAllConstants
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
//==========================================================================
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
void ZCCCompiler::CopyConstants(TArray<ZCC_ConstantWork> &dest, TArray<ZCC_ConstantDef*> &Constants, PSymbolTable *ot)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
for (auto c : Constants)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
dest.Push({ c, ot });
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
}
}
//==========================================================================
//
// ZCCCompiler :: CompileAllConstants
//
// Make symbols from every constant defined at all levels.
// Since constants may only depend on other constants this can be done
// without any more involved processing of the AST as a first step.
//
//==========================================================================
void ZCCCompiler::CompileAllConstants()
{
// put all constants in one list to make resolving this easier.
TArray<ZCC_ConstantWork> constantwork;
- renamed the symbol tables holding the tree nodes so that the code becomes clearer to read. Necessary because the unclear naming caused: - fixed: The tree nodes for classes and struct members were stored in the global tree nodes table. - sort variable declarations into their own list for processing.
2016-10-09 07:09:17 +00:00
CopyConstants(constantwork, Constants, OutputSymbols);
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
for (auto c : Classes)
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
CopyConstants(constantwork, c->Constants, &c->Type()->Symbols);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
}
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
for (auto s : Structs)
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
CopyConstants(constantwork, s->Constants, &s->Type()->Symbols);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
}
// Before starting to resolve the list, let's create symbols for all already resolved ones first (i.e. all literal constants), to reduce work.
for (unsigned i = 0; i<constantwork.Size(); i++)
{
if (constantwork[i].node->Value->NodeType == AST_ExprConstant)
{
AddConstant(constantwork[i]);
// Remove the constant from the list
constantwork.Delete(i);
i--;
}
}
bool donesomething = true;
// Now go through this list until no more constants can be resolved. The remaining ones will be non-constant values.
while (donesomething && constantwork.Size() > 0)
{
donesomething = false;
for (unsigned i = 0; i < constantwork.Size(); i++)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (CompileConstant(constantwork[i].node, constantwork[i].outputtable))
{
AddConstant(constantwork[i]);
// Remove the constant from the list
constantwork.Delete(i);
i--;
donesomething = true;
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
for (unsigned i = 0; i < constantwork.Size(); i++)
{
Error(constantwork[i].node, "%s is not a constant", FName(constantwork[i].node->NodeName).GetChars());
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
//==========================================================================
//
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
// ZCCCompiler :: AddConstant
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
// Adds a constant to its assigned symbol table
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
void ZCCCompiler::AddConstant(ZCC_ConstantWork &constant)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
auto def = constant.node;
auto val = def->Value;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
if (val->NodeType == AST_ExprConstant)
{
ZCC_ExprConstant *cval = static_cast<ZCC_ExprConstant *>(val);
if (cval->Type == TypeString)
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
def->Symbol = new PSymbolConstString(def->NodeName, *(cval->StringVal));
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
else if (cval->Type->IsA(RUNTIME_CLASS(PInt)))
{
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// How do we get an Enum type in here without screwing everything up???
//auto type = def->Type != nullptr ? def->Type : cval->Type;
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
def->Symbol = new PSymbolConstNumeric(def->NodeName, cval->Type, cval->IntVal);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
else if (cval->Type->IsA(RUNTIME_CLASS(PFloat)))
{
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
if (def->Type != nullptr)
{
Error(def, "Enum members must be integer values");
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
def->Symbol = new PSymbolConstNumeric(def->NodeName, cval->Type, cval->DoubleVal);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
else
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
Error(def->Value, "Bad type for constant definiton");
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
def->Symbol = nullptr;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (def->Symbol == nullptr)
{
// Create a dummy constant so we don't make any undefined value warnings.
def->Symbol = new PSymbolConstNumeric(def->NodeName, TypeError, 0);
}
constant.outputtable->ReplaceSymbol(def->Symbol);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
}
//==========================================================================
//
// ZCCCompiler :: CompileConstant
//
// For every constant definition, evaluate its value (which should result
// in a constant), and create a symbol for it.
//
//==========================================================================
bool ZCCCompiler::CompileConstant(ZCC_ConstantDef *def, PSymbolTable *sym)
{
assert(def->Symbol == nullptr);
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
ZCC_Expression *val = Simplify(def->Value, sym, true);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
def->Value = val;
return (val->NodeType == AST_ExprConstant);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
//==========================================================================
//
// ZCCCompiler :: Simplify
//
// For an expression,
// Evaluate operators whose arguments are both constants, replacing it
// with a new constant.
// For a binary operator with one constant argument, put it on the right-
// hand operand, where permitted.
// Perform automatic type promotion.
//
//==========================================================================
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
ZCC_Expression *ZCCCompiler::Simplify(ZCC_Expression *root, PSymbolTable *sym, bool wantconstant)
{
SimplifyingConstant = wantconstant;
return DoSimplify(root, sym);
}
ZCC_Expression *ZCCCompiler::DoSimplify(ZCC_Expression *root, PSymbolTable *sym)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
Remove IsUnary() and IsBinary() - This information is already stored in the node's NodeType field, so there's no reason to go do a table lookup for it elsewhere. Must have been a brain fart when I wrote them in the first place.
2013-11-02 02:28:00 +00:00
if (root->NodeType == AST_ExprUnary)
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return SimplifyUnary(static_cast<ZCC_ExprUnary *>(root), sym);
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
}
Remove IsUnary() and IsBinary() - This information is already stored in the node's NodeType field, so there's no reason to go do a table lookup for it elsewhere. Must have been a brain fart when I wrote them in the first place.
2013-11-02 02:28:00 +00:00
else if (root->NodeType == AST_ExprBinary)
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return SimplifyBinary(static_cast<ZCC_ExprBinary *>(root), sym);
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
}
else if (root->Operation == PEX_ID)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return IdentifyIdentifier(static_cast<ZCC_ExprID *>(root), sym);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
else if (root->Operation == PEX_MemberAccess)
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return SimplifyMemberAccess(static_cast<ZCC_ExprMemberAccess *>(root), sym);
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
}
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
else if (root->Operation == PEX_FuncCall)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return SimplifyFunctionCall(static_cast<ZCC_ExprFuncCall *>(root), sym);
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
return root;
}
//==========================================================================
//
// ZCCCompiler :: SimplifyUnary
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *ZCCCompiler::SimplifyUnary(ZCC_ExprUnary *unary, PSymbolTable *sym)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
unary->Operand = DoSimplify(unary->Operand, sym);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (unary->Operand->Type == nullptr)
{
return unary;
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
ZCC_OpProto *op = PromoteUnary(unary->Operation, unary->Operand);
if (op == NULL)
{ // Oh, poo!
unary->Type = TypeError;
}
else if (unary->Operand->Operation == PEX_ConstValue)
{
return op->EvalConst1(static_cast<ZCC_ExprConstant *>(unary->Operand));
}
return unary;
}
//==========================================================================
//
// ZCCCompiler :: SimplifyBinary
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *ZCCCompiler::SimplifyBinary(ZCC_ExprBinary *binary, PSymbolTable *sym)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
binary->Left = DoSimplify(binary->Left, sym);
binary->Right = DoSimplify(binary->Right, sym);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (binary->Left->Type == nullptr || binary->Right->Type == nullptr)
{
// We do not know yet what this is so we cannot promote it (yet.)
return binary;
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
ZCC_OpProto *op = PromoteBinary(binary->Operation, binary->Left, binary->Right);
if (op == NULL)
{
binary->Type = TypeError;
}
else if (binary->Left->Operation == PEX_ConstValue &&
binary->Right->Operation == PEX_ConstValue)
{
return op->EvalConst2(static_cast<ZCC_ExprConstant *>(binary->Left),
static_cast<ZCC_ExprConstant *>(binary->Right), AST.Strings);
}
return binary;
}
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
//==========================================================================
//
// ZCCCompiler :: SimplifyMemberAccess
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *ZCCCompiler::SimplifyMemberAccess(ZCC_ExprMemberAccess *dotop, PSymbolTable *symt)
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
PSymbolTable *symtable;
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
// TBD: Is it safe to simplify the left side here when not processing a constant?
dotop->Left = DoSimplify(dotop->Left, symt);
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
if (dotop->Left->Operation == PEX_TypeRef)
{ // Type refs can be evaluated now.
PType *ref = static_cast<ZCC_ExprTypeRef *>(dotop->Left)->RefType;
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
PSymbol *sym = ref->Symbols.FindSymbolInTable(dotop->Right, symtable);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (sym != nullptr)
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
ZCC_Expression *expr = NodeFromSymbol(sym, dotop, symtable);
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
if (expr != nullptr)
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return expr;
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
}
}
else if (dotop->Left->Operation == PEX_Super)
{
symt = symt->GetParentTable();
if (symt != nullptr)
{
PSymbol *sym = symt->FindSymbolInTable(dotop->Right, symtable);
if (sym != nullptr)
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *expr = NodeFromSymbol(sym, dotop, symtable);
if (expr != nullptr)
{
return expr;
}
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
}
}
}
return dotop;
}
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
//==========================================================================
//
// ZCCCompiler :: SimplifyFunctionCall
//
// This may replace a function call with cast(s), since they look like the
// same thing to the parser.
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *ZCCCompiler::SimplifyFunctionCall(ZCC_ExprFuncCall *callop, PSymbolTable *sym)
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
{
ZCC_FuncParm *parm;
int parmcount = 0;
parm = callop->Parameters;
if (parm != NULL)
{
do
{
parmcount++;
assert(parm->NodeType == AST_FuncParm);
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
parm->Value = DoSimplify(parm->Value, sym);
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
parm = static_cast<ZCC_FuncParm *>(parm->SiblingNext);
}
while (parm != callop->Parameters);
}
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
// Only simplify the 'function' part if we want to retrieve a constant.
// This is necessary to evaluate the type casts, but for actual functions
// the simplification process is destructive and has to be avoided.
if (SimplifyingConstant)
{
callop->Function = DoSimplify(callop->Function, sym);
}
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
// If the left side is a type ref, then this is actually a cast
// and not a function call.
if (callop->Function->Operation == PEX_TypeRef)
{
if (parmcount != 1)
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
Error(callop, "Type cast requires one parameter");
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
callop->ToErrorNode();
}
else
{
PType *dest = static_cast<ZCC_ExprTypeRef *>(callop->Function)->RefType;
const PType::Conversion *route[CONVERSION_ROUTE_SIZE];
int routelen = parm->Value->Type->FindConversion(dest, route, countof(route));
if (routelen < 0)
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
///FIXME: Need real type names
- changed order of script files to match the old DECORATE list for easy comparison of the disassembly. - added a descriptive name to all types for error messages. - added a generic type cast node to the code generator. - added a few more cast operations to the 'cast' VM instruction. - extended FxClassTypeCast to handle all possible input that can be cast to a class pointer, not just names.
2016-10-16 17:42:22 +00:00
Error(callop, "Cannot convert %s to %s", parm->Value->Type->DescriptiveName(), dest->DescriptiveName());
Interpret function calls to type refs as type casts
2013-10-31 01:53:02 +00:00
callop->ToErrorNode();
}
else
{
ZCC_Expression *val = ApplyConversion(parm->Value, route, routelen);
assert(val->Type == dest);
return val;
}
}
}
return callop;
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//==========================================================================
//
// ZCCCompiler :: PromoteUnary
//
// Converts the operand into a format preferred by the operator.
//
//==========================================================================
ZCC_OpProto *ZCCCompiler::PromoteUnary(EZCCExprType op, ZCC_Expression *&expr)
{
if (expr->Type == TypeError)
{
return NULL;
}
const PType::Conversion *route[CONVERSION_ROUTE_SIZE];
int routelen = countof(route);
ZCC_OpProto *proto = ZCC_OpInfo[op].FindBestProto(expr->Type, route, routelen);
if (proto != NULL)
{
expr = ApplyConversion(expr, route, routelen);
}
return proto;
}
//==========================================================================
//
// ZCCCompiler :: PromoteBinary
//
// Converts the operands into a format (hopefully) compatible with the
// operator.
//
//==========================================================================
ZCC_OpProto *ZCCCompiler::PromoteBinary(EZCCExprType op, ZCC_Expression *&left, ZCC_Expression *&right)
{
// If either operand is of type 'error', the result is also 'error'
if (left->Type == TypeError || right->Type == TypeError)
{
return NULL;
}
const PType::Conversion *route1[CONVERSION_ROUTE_SIZE], *route2[CONVERSION_ROUTE_SIZE];
int route1len = countof(route1), route2len = countof(route2);
ZCC_OpProto *proto = ZCC_OpInfo[op].FindBestProto(left->Type, route1, route1len, right->Type, route2, route2len);
if (proto != NULL)
{
left = ApplyConversion(left, route1, route1len);
right = ApplyConversion(right, route2, route2len);
}
return proto;
}
//==========================================================================
//
// ZCCCompiler :: ApplyConversion
//
//==========================================================================
ZCC_Expression *ZCCCompiler::ApplyConversion(ZCC_Expression *expr, const PType::Conversion **route, int routelen)
{
for (int i = 0; i < routelen; ++i)
{
if (expr->Operation != PEX_ConstValue)
{
expr = AddCastNode(route[i]->TargetType, expr);
}
else
{
route[i]->ConvertConstant(static_cast<ZCC_ExprConstant *>(expr), AST.Strings);
}
}
return expr;
}
//==========================================================================
//
// ZCCCompiler :: AddCastNode
//
//==========================================================================
ZCC_Expression *ZCCCompiler::AddCastNode(PType *type, ZCC_Expression *expr)
{
assert(expr->Operation != PEX_ConstValue && "Expression must not be constant");
// TODO: add a node here
return expr;
}
//==========================================================================
//
// ZCCCompiler :: IdentifyIdentifier
//
// Returns a node that represents what the identifer stands for.
//
//==========================================================================
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_Expression *ZCCCompiler::IdentifyIdentifier(ZCC_ExprID *idnode, PSymbolTable *symt)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
// Check the symbol table for the identifier.
PSymbolTable *table;
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
PSymbol *sym = symt->FindSymbolInTable(idnode->Identifier, table);
// GlobalSymbols cannot be the parent of a class's symbol table so we have to look for global symbols explicitly.
if (sym == nullptr && symt != &GlobalSymbols) sym = GlobalSymbols.FindSymbolInTable(idnode->Identifier, table);
if (sym != nullptr)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
ZCC_Expression *node = NodeFromSymbol(sym, idnode, table);
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
if (node != NULL)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
return node;
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
}
}
- when simplifying a non-constant expression, leave any identifier that cannot be trivially simplified to a constant alone. It is better to defer resolution to more context-aware code for those.
2016-10-16 07:31:03 +00:00
else if (SimplifyingConstant) // leave unknown identifiers alone when simplifying non-constants. It is impossible to know what they are here.
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
// Also handle line specials.
// To call this like a function this needs to be done differently, but for resolving constants this is ok.
int spec = P_FindLineSpecial(FName(idnode->Identifier).GetChars());
if (spec != 0)
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
ZCC_ExprConstant *val = static_cast<ZCC_ExprConstant *>(AST.InitNode(sizeof(*val), AST_ExprConstant, idnode));
val->Operation = PEX_ConstValue;
val->Type = TypeSInt32;
val->IntVal = spec;
return val;
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
Error(idnode, "Unknown identifier '%s'", FName(idnode->Identifier).GetChars());
idnode->ToErrorNode();
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
return idnode;
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
}
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//==========================================================================
//
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
// ZCCCompiler :: NodeFromSymbol
//
//==========================================================================
Store known but uncompiled nodes in the symbol table - Don't bother keeping track of uncompiled nodes in a special table. Use the regular symbol table instead. This should in the future make compiling nodes referenced deeper than (and before) their definitions fairly straightforward. - Also, break up the compiler's Message() function into Warn() and Error() and get rid of zcc_errors.h. I can't really see having a set of error numbers being useful.
2016-03-13 03:22:29 +00:00
ZCC_Expression *ZCCCompiler::NodeFromSymbol(PSymbol *sym, ZCC_Expression *source, PSymbolTable *table)
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
{
- implemented complete resolving of constants - both global and class-local. This uses a different algorithm as the old implementation - instead of recursively resolving unknown symbols it will first collect all constants from all scopes and then process them in one operation, doing multiple passes over the list until no more constants can be resolved anymore.
2016-10-08 20:16:10 +00:00
assert(sym != nullptr);
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
if (sym->IsKindOf(RUNTIME_CLASS(PSymbolConst)))
{
return NodeFromSymbolConst(static_cast<PSymbolConst *>(sym), source);
}
else if (sym->IsKindOf(RUNTIME_CLASS(PSymbolType)))
{
return NodeFromSymbolType(static_cast<PSymbolType *>(sym), source);
}
return NULL;
}
//==========================================================================
//
// ZCCCompiler :: NodeFromSymbolConst
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
//
// Returns a new AST constant node with the symbol's content.
//
//==========================================================================
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
ZCC_ExprConstant *ZCCCompiler::NodeFromSymbolConst(PSymbolConst *sym, ZCC_Expression *idnode)
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
ZCC_ExprConstant *val = static_cast<ZCC_ExprConstant *>(AST.InitNode(sizeof(*val), AST_ExprConstant, idnode));
val->Operation = PEX_ConstValue;
Allow using constants in other constants before their definitions. - Something like this is now valid: const foo = bar + 10; const bar = 1000;
2013-10-26 03:01:18 +00:00
if (sym == NULL)
{
val->Type = TypeError;
val->IntVal = 0;
}
else if (sym->IsKindOf(RUNTIME_CLASS(PSymbolConstString)))
Add evaluation of constant unary and binary expressions - Added ZCCCompiler class as a place to generate IR and symbols from an AST. Right now, all it does is simplify constant expressions into constant values. - Do type promotion on the AST where appropriate. - Added true and false tokens to the parser driver.
2013-10-24 04:06:32 +00:00
{
val->StringVal = AST.Strings.Alloc(static_cast<PSymbolConstString *>(sym)->Str);
val->Type = TypeString;
}
else
{
val->Type = sym->ValueType;
if (val->Type != TypeError)
{
assert(sym->IsKindOf(RUNTIME_CLASS(PSymbolConstNumeric)));
if (sym->ValueType->IsKindOf(RUNTIME_CLASS(PInt)))
{
val->IntVal = static_cast<PSymbolConstNumeric *>(sym)->Value;
}
else
{
assert(sym->ValueType->IsKindOf(RUNTIME_CLASS(PFloat)));
val->DoubleVal = static_cast<PSymbolConstNumeric *>(sym)->Float;
}
}
}
return val;
}
Allow access to the numeric properties. - Identifiers can now evaluate to type references. - The dot operator can now find symbols in type references.
2013-10-30 02:55:32 +00:00
//==========================================================================
//
// ZCCCompiler :: NodeFromSymbolType
//
// Returns a new AST type ref node with the symbol's content.
//
//==========================================================================
ZCC_ExprTypeRef *ZCCCompiler::NodeFromSymbolType(PSymbolType *sym, ZCC_Expression *idnode)
{
ZCC_ExprTypeRef *ref = static_cast<ZCC_ExprTypeRef *>(AST.InitNode(sizeof(*ref), AST_ExprTypeRef, idnode));
ref->Operation = PEX_TypeRef;
ref->RefType = sym->Type;
ref->Type = NewClassPointer(RUNTIME_CLASS(PType));
return ref;
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
//==========================================================================
//
// ZCCCompiler :: CompileAllFields
//
// builds the internal structure of all classes and structs
//
//==========================================================================
void ZCCCompiler::CompileAllFields()
{
// Create copies of the arrays which can be altered
auto Classes = this->Classes;
auto Structs = this->Structs;
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
TMap<PClass*, bool> HasNativeChildren;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// first step: Look for native classes with native children.
// These may not have any variables added to them because it'd clash with the native definitions.
for (unsigned i = 0; i < Classes.Size(); i++)
{
auto c = Classes[i];
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (c->Type()->Size != TentativeClass && c->Fields.Size() > 0)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
// We need to search the global class table here because not all children may have a scripted definition attached.
for (auto ac : PClass::AllClasses)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (ac->ParentClass == c->Type() && ac->Size != TentativeClass)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
// Only set a marker here, so that we can print a better message when the actual fields get added.
HasNativeChildren.Insert(ac, true);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
}
}
}
}
bool donesomething = true;
while (donesomething && (Structs.Size() > 0 || Classes.Size() > 0))
{
donesomething = false;
for (unsigned i = 0; i < Structs.Size(); i++)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (CompileFields(Structs[i]->Type(), Structs[i]->Fields, Structs[i]->Outer, &Structs[i]->TreeNodes, true))
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
// Remove from the list if all fields got compiled.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Structs.Delete(i--);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
donesomething = true;
}
}
for (unsigned i = 0; i < Classes.Size(); i++)
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto type = Classes[i]->Type();
if (type->Size == TentativeClass)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (type->ParentClass->Size == TentativeClass)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
// we do not know the parent class's size yet, so skip this class for now.
continue;
}
else
{
// Inherit the size of the parent class
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
type->Size = Classes[i]->Type()->ParentClass->Size;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
}
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
if (CompileFields(type, Classes[i]->Fields, nullptr, &Classes[i]->TreeNodes, false, !!HasNativeChildren.CheckKey(type)))
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
// Remove from the list if all fields got compiled.
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Classes.Delete(i--);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
donesomething = true;
}
}
}
- a bit of cleanup.
2016-10-09 20:01:23 +00:00
// This really should never happen, but if it does, let's better print an error.
for (auto s : Structs)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(s->strct, "Unable to resolve all fields for struct %s", FName(s->NodeName()).GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- a bit of cleanup.
2016-10-09 20:01:23 +00:00
for (auto s : Classes)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
Error(s->cls, "Unable to resolve all fields for class %s", FName(s->NodeName()).GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
}
//==========================================================================
//
// ZCCCompiler :: CompileFields
//
// builds the internal structure of a single class or struct
//
//==========================================================================
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
bool ZCCCompiler::CompileFields(PStruct *type, TArray<ZCC_VarDeclarator *> &Fields, PClass *Outer, PSymbolTable *TreeNodes, bool forstruct, bool hasnativechildren)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
while (Fields.Size() > 0)
{
auto field = Fields[0];
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
FieldDesc *fd = nullptr;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
PType *fieldtype = DetermineType(type, field, field->Names->Name, field->Type, true, true);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// For structs only allow 'deprecated', for classes exclude function qualifiers.
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
int notallowed = forstruct?
ZCC_Latent | ZCC_Final | ZCC_Action | ZCC_Static | ZCC_FuncConst | ZCC_Abstract | ZCC_Virtual | ZCC_Override | ZCC_Meta | ZCC_Extension :
ZCC_Latent | ZCC_Final | ZCC_Action | ZCC_Static | ZCC_FuncConst | ZCC_Abstract | ZCC_Virtual | ZCC_Override | ZCC_Extension;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
if (field->Flags & notallowed)
{
- Fixed Clang compiler errors and some warnings.
2016-10-25 14:53:14 +00:00
Error(field, "Invalid qualifiers for %s (%s not allowed)", FName(field->Names->Name).GetChars(), FlagsToString(field->Flags & notallowed).GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
field->Flags &= notallowed;
}
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
uint32_t varflags = 0;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// These map directly to implementation flags.
if (field->Flags & ZCC_Private) varflags |= VARF_Private;
if (field->Flags & ZCC_Protected) varflags |= VARF_Protected;
if (field->Flags & ZCC_Deprecated) varflags |= VARF_Deprecated;
if (field->Flags & ZCC_ReadOnly) varflags |= VARF_ReadOnly;
if (field->Flags & ZCC_Native)
{
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
varflags |= VARF_Native | VARF_Transient;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
if (field->Flags & ZCC_Meta)
{
- scriptified Heretic's blaster. - scriptified all Effect functions of Fastprojectile's children - implemented access to class meta data. - added a VM instruction to retrieve the class metadata, to eliminate the overhead of the function call that would otherwise be needed. - made GetClass() a builtin so that it can use the new instruction Important note about this commit: Scriptifying CFlameMissile::Effect revealed a problem with the virtual function interface: In order to work, this needs to be explicitly enabled for each single native class that may be used as a base for a scripted class. Needless to say, this will end up way too much work, as there are over 100 native classes, excluding those which will be scriptified. But in order to fix the problem this partially broken state needs to be committed first.
2016-11-24 19:02:44 +00:00
varflags |= VARF_Static|VARF_ReadOnly; // metadata implies readonly
if (!(field->Flags & ZCC_Native))
{
// Non-native meta data is not implemented yet and requires some groundwork in the class copy code.
Error(field, "Metadata member %s must be native", FName(field->Names->Name).GetChars());
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
if (field->Type->ArraySize != nullptr)
{
fieldtype = ResolveArraySize(fieldtype, field->Type->ArraySize, &type->Symbols);
}
auto name = field->Names;
do
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
if (AddTreeNode(name->Name, name, TreeNodes, !forstruct))
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
auto thisfieldtype = fieldtype;
if (name->ArraySize != nullptr)
{
thisfieldtype = ResolveArraySize(thisfieldtype, name->ArraySize, &type->Symbols);
}
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
if (varflags & VARF_Native)
{
- scriptified Heretic's blaster. - scriptified all Effect functions of Fastprojectile's children - implemented access to class meta data. - added a VM instruction to retrieve the class metadata, to eliminate the overhead of the function call that would otherwise be needed. - made GetClass() a builtin so that it can use the new instruction Important note about this commit: Scriptifying CFlameMissile::Effect revealed a problem with the virtual function interface: In order to work, this needs to be explicitly enabled for each single native class that may be used as a base for a scripted class. Needless to say, this will end up way too much work, as there are over 100 native classes, excluding those which will be scriptified. But in order to fix the problem this partially broken state needs to be committed first.
2016-11-24 19:02:44 +00:00
auto querytype = (varflags & VARF_Static) ? type->GetClass() : type;
fd = FindField(querytype, FName(name->Name).GetChars());
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
if (fd == nullptr)
{
Error(field, "The member variable '%s.%s' has not been exported from the executable.", type->TypeName.GetChars(), FName(name->Name).GetChars());
}
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
else if (thisfieldtype->Size != fd->FieldSize && fd->BitValue == 0)
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
{
Error(field, "The member variable '%s.%s' has mismatching sizes in internal and external declaration. (Internal = %d, External = %d)", type->TypeName.GetChars(), FName(name->Name).GetChars(), fd->FieldSize, thisfieldtype->Size);
}
// Q: Should we check alignment, too? A mismatch may be an indicator for bad assumptions.
else
{
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
// for bit fields the type must point to the source variable.
if (fd->BitValue != 0) thisfieldtype = fd->FieldSize == 1 ? TypeUInt8 : fd->FieldSize == 2 ? TypeUInt16 : TypeUInt32;
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
type->AddNativeField(name->Name, thisfieldtype, fd->FieldOffset, varflags, fd->BitValue);
}
}
else if (hasnativechildren)
{
Error(field, "Cannot add field %s to %s. %s has native children which means it size may not change.", type->TypeName.GetChars(), fd->FieldSize, FName(name->Name).GetChars());
}
else
{
type->AddField(name->Name, thisfieldtype, varflags);
}
- changed PClass:AddField so that it doesn't create defaults if none are present. The ZScript compiler will not create any defaults until after all fields have been added and this function created bad data if it was run on a class without defaults. - added checks for duplicate field names. - moved the tree node symbol tables out of PSymbolTreeNode to the worker data. That symbol is a bad location because it restricts the usefulness of the symbol class which is also needed for variables which use different AST structs. - fixed some memory management issues with the work classes for the compiler that became apparent after moving the symbol tables in there. In several places these were copied around, possibly losing data.
2016-10-09 23:18:47 +00:00
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
name = static_cast<ZCC_VarName*>(name->SiblingNext);
} while (name != field->Names);
Fields.Delete(0);
}
return Fields.Size() == 0;
}
//==========================================================================
//
// ZCCCompiler :: FieldFlagsToString
//
// creates a string for a field's flags
//
//==========================================================================
FString ZCCCompiler::FlagsToString(uint32_t flags)
{
const char *flagnames[] = { "native", "static", "private", "protected", "latent", "final", "meta", "action", "deprecated", "readonly", "funcconst", "abstract" };
FString build;
for (int i = 0; i < 12; i++)
{
if (flags & (1 << i))
{
if (build.IsNotEmpty()) build += ", ";
build += flagnames[i];
}
}
return build;
}
//==========================================================================
//
// ZCCCompiler :: DetermineType
//
// retrieves the type for this field, for arrays the type of a single entry.
//
//==========================================================================
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
PType *ZCCCompiler::DetermineType(PType *outertype, ZCC_TreeNode *field, FName name, ZCC_Type *ztype, bool allowarraytypes, bool formember)
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
PType *retval = TypeError;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
if (!allowarraytypes && ztype->ArraySize != nullptr)
{
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
Error(field, "%s: Array type not allowed", name.GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
return TypeError;
}
switch (ztype->NodeType)
{
case AST_BasicType:
{
auto btype = static_cast<ZCC_BasicType *>(ztype);
switch (btype->Type)
{
case ZCC_SInt8:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
retval = formember? TypeSInt8 : (PType*)TypeError;
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_UInt8:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
retval = formember ? TypeUInt8 : (PType*)TypeError;
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_SInt16:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
retval = formember ? TypeSInt16 : (PType*)TypeError;
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_UInt16:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
retval = formember ? TypeUInt16 : (PType*)TypeError;
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_SInt32:
case ZCC_IntAuto: // todo: for enums, autoselect appropriately sized int
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeSInt32;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_UInt32:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeUInt32;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_Bool:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeBool;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_FloatAuto:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
retval = formember ? TypeFloat32 : TypeFloat64;
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
break;
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_Float64:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeFloat64;
- changed order of script files to match the old DECORATE list for easy comparison of the disassembly. - added a descriptive name to all types for error messages. - added a generic type cast node to the code generator. - added a few more cast operations to the 'cast' VM instruction. - extended FxClassTypeCast to handle all possible input that can be cast to a class pointer, not just names.
2016-10-16 17:42:22 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_String:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeString;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_Name:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeName;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_Vector2:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeVector2;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_Vector3:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeVector3;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
case ZCC_State:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeState;
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
case ZCC_Color:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeColor;
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
case ZCC_Sound:
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = TypeSound;
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- added automatic type deduction capabilities for local variables. If you type 'let variable = value;' the type of 'variable' will be deduced from the given value. This is mostly of interest for type casting pointers, because 'let p = Classtype(objectvar);' does not require writing the class type name twice.
2016-12-05 12:24:42 +00:00
case ZCC_Let:
retval = TypeAuto;
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
case ZCC_UserType:
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
// statelabel et.al. are not tokens - there really is no need to, it works just as well as an identifier. Maybe the same should be done for some other types, too?
switch (btype->UserType->Id)
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
{
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
case NAME_StateLabel:
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
retval = TypeStateLabel;
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
break;
case NAME_SpriteID:
retval = TypeSpriteID;
break;
case NAME_TextureID:
retval = TypeTextureID;
break;
default:
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
retval = ResolveUserType(btype, &outertype->Symbols);
- added two new integral types SpriteID and TextureID, both are needed to allow proper serialization as they require something different to be written out than a regular integer.
2016-11-21 12:45:33 +00:00
break;
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
break;
- Fixed more GCC/Clang warnings.
2016-11-26 11:45:17 +00:00
default:
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
case AST_MapType:
if (allowarraytypes)
{
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
Error(field, "%s: Map types not implemented yet", name.GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
// Todo: Decide what we allow here and if it makes sense to allow more complex constructs.
auto mtype = static_cast<ZCC_MapType *>(ztype);
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = NewMap(DetermineType(outertype, field, name, mtype->KeyType, false, false), DetermineType(outertype, field, name, mtype->ValueType, false, false));
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
break;
case AST_DynArrayType:
if (allowarraytypes)
{
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
Error(field, "%s: Dynamic array types not implemented yet", name.GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
auto atype = static_cast<ZCC_DynArrayType *>(ztype);
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = NewDynArray(DetermineType(outertype, field, name, atype->ElementType, false, false));
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
break;
case AST_ClassType:
{
auto ctype = static_cast<ZCC_ClassType *>(ztype);
if (ctype->Restriction == nullptr)
{
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = NewClassPointer(RUNTIME_CLASS(DObject));
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
else
{
auto sym = outertype->Symbols.FindSymbol(ctype->Restriction->Id, true);
if (sym == nullptr) sym = GlobalSymbols.FindSymbol(ctype->Restriction->Id, false);
if (sym == nullptr)
{
Error(field, "%s: Unknown identifier", FName(ctype->Restriction->Id).GetChars());
return TypeError;
}
auto typesym = dyn_cast<PSymbolType>(sym);
if (typesym == nullptr || !typesym->Type->IsKindOf(RUNTIME_CLASS(PClass)))
{
Error(field, "%s does not represent a class type", FName(ctype->Restriction->Id).GetChars());
return TypeError;
}
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
retval = NewClassPointer(static_cast<PClass *>(typesym->Type));
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- Fixed more GCC/Clang warnings.
2016-11-26 11:45:17 +00:00
default:
break;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
if (retval != TypeError && retval->MemberOnly && !formember)
{
- changed GetStore/LoadMoveOp and GetRegType to be based on a set of member variables instead of using virtual functions to return such trivial values. This not only creates significantly better code but also allows overriding these settings without having to create a new class just for changing them (as is needed for vectors.)
2016-10-28 08:44:01 +00:00
Error(field, "Invalid type %s", retval->DescriptiveName());
- restrict 8 and 16 bit integer types to struct and class members - and I am not really sure if they should even be allowed there, aside from legacy readonly properties. For 32 bit floats the same would hold, but those are already mapped to 64 bit doubles silently. ZDoom only uses these types in a very few isolated places, and even those can be removed without problems, so it's very doubtful that having support for these types is of any benefit - on the other hand, having them will most likely introduce more code than is saved in the data by using them...
2016-10-16 10:47:26 +00:00
return TypeError;
}
return retval;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
//==========================================================================
//
// ZCCCompiler :: ResolveUserType
//
// resolves a user type and returns a matching PType
//
//==========================================================================
PType *ZCCCompiler::ResolveUserType(ZCC_BasicType *type, PSymbolTable *symt)
{
// Check the symbol table for the identifier.
PSymbolTable *table;
PSymbol *sym = symt->FindSymbolInTable(type->UserType->Id, table);
// GlobalSymbols cannot be the parent of a class's symbol table so we have to look for global symbols explicitly.
if (sym == nullptr && symt != &GlobalSymbols) sym = GlobalSymbols.FindSymbolInTable(type->UserType->Id, table);
if (sym != nullptr && sym->IsKindOf(RUNTIME_CLASS(PSymbolType)))
{
- added readonly pointers. They need to be defined with 'readonly<classtype>'. These are significantly different from declaring a field readonly in that they do not disallow modification of the variable itself but what it points to. For the actor defaults this is necessary to prevent accidental modification. A readonly pointer is actually a different type than a regular pointer. - fixed code generation for dynamic cast. It was missing the jump instruction after the compare.
2016-11-05 12:51:46 +00:00
auto ptype = static_cast<PSymbolType *>(sym)->Type;
if (ptype->IsKindOf(RUNTIME_CLASS(PEnum)))
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
{
return TypeSInt32; // hack this to an integer until we can resolve the enum mess.
}
- made all internal struct types NativeStructs so that all variables of their types are made references.
2016-11-21 18:20:27 +00:00
if (ptype->IsKindOf(RUNTIME_CLASS(PNativeStruct))) // native structs and classes cannot be instantiated, they always get used as reference.
- added null pointer. - fixed: FxMemberIdentifier checked for ClassPointers instead of object pointers to resolve the left hand side of the expression. - allow comparison of pointers.
2016-10-21 22:50:04 +00:00
{
- added readonly pointers. They need to be defined with 'readonly<classtype>'. These are significantly different from declaring a field readonly in that they do not disallow modification of the variable itself but what it points to. For the actor defaults this is necessary to prevent accidental modification. A readonly pointer is actually a different type than a regular pointer. - fixed code generation for dynamic cast. It was missing the jump instruction after the compare.
2016-11-05 12:51:46 +00:00
return NewPointer(ptype, type->isconst);
- added null pointer. - fixed: FxMemberIdentifier checked for ClassPointers instead of object pointers to resolve the left hand side of the expression. - allow comparison of pointers.
2016-10-21 22:50:04 +00:00
}
- added readonly pointers. They need to be defined with 'readonly<classtype>'. These are significantly different from declaring a field readonly in that they do not disallow modification of the variable itself but what it points to. For the actor defaults this is necessary to prevent accidental modification. A readonly pointer is actually a different type than a regular pointer. - fixed code generation for dynamic cast. It was missing the jump instruction after the compare.
2016-11-05 12:51:46 +00:00
return ptype;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
}
- scriptified A_Mushroom to test something a bit more complex. - fixed: FxMinusSign trashed local variables that were used with negation. - fixed: FxConditional only handled ints and floats, but not pointers and strings. - fixed: A 'no states in non-actors' error was triggered, even for classes without any states.
2016-10-30 13:00:11 +00:00
Error(type, "Unable to resolve %s as type.", FName(type->UserType->Id).GetChars());
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
return TypeError;
}
//==========================================================================
//
// ZCCCompiler :: ResolveArraySize
//
// resolves the array size and returns a matching type.
//
//==========================================================================
PType *ZCCCompiler::ResolveArraySize(PType *baseType, ZCC_Expression *arraysize, PSymbolTable *sym)
{
- fixed: Type determination of multi-dimensional arrays failed, apparently because ZCCCompiler::Simplify is too broken to leave the node's ring list intact. Instead of trying to fix Simplify, which seems to be a lost cause, the ring list now gets unraveled into an array which is immune from this type of problem.
2016-12-09 22:00:34 +00:00
TArray<ZCC_Expression *> indices;
// Simplify is too broken to resolve this inside the ring list so unravel the list into an array before starting to simplify its components.
auto node = arraysize;
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
do
{
- fixed: Type determination of multi-dimensional arrays failed, apparently because ZCCCompiler::Simplify is too broken to leave the node's ring list intact. Instead of trying to fix Simplify, which seems to be a lost cause, the ring list now gets unraveled into an array which is immune from this type of problem.
2016-12-09 22:00:34 +00:00
indices.Push(node);
node = static_cast<ZCC_Expression*>(node->SiblingNext);
} while (node != arraysize);
for (auto node : indices)
{
auto val = Simplify(node, sym, true);
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
if (val->Operation != PEX_ConstValue || !val->Type->IsA(RUNTIME_CLASS(PInt)))
{
Error(arraysize, "Array index must be an integer constant");
return TypeError;
}
int size = static_cast<ZCC_ExprConstant *>(val)->IntVal;
if (size < 1)
{
Error(arraysize, "Array size must be positive");
return TypeError;
}
baseType = NewArray(baseType, size);
- fixed: Type determination of multi-dimensional arrays failed, apparently because ZCCCompiler::Simplify is too broken to leave the node's ring list intact. Instead of trying to fix Simplify, which seems to be a lost cause, the ring list now gets unraveled into an array which is immune from this type of problem.
2016-12-09 22:00:34 +00:00
}
- parse the list of class/struct fields and add the PField entries to their type. - fixed a few issues with trying to access a class's Type before it got initialized.
2016-10-09 13:47:31 +00:00
return baseType;
}
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
//==========================================================================
//
// ZCCCompiler :: GetInt - Input must be a constant expression
//
//==========================================================================
int ZCCCompiler::GetInt(ZCC_Expression *expr)
{
if (expr->Type == TypeError)
{
return 0;
}
const PType::Conversion *route[CONVERSION_ROUTE_SIZE];
int routelen = expr->Type->FindConversion(TypeSInt32, route, countof(route));
if (routelen < 0)
{
Error(expr, "Cannot convert to integer");
return 0;
}
else
{
if (expr->Type->IsKindOf(RUNTIME_CLASS(PFloat)))
{
Warn(expr, "Truncation of floating point value");
}
auto ex = static_cast<ZCC_ExprConstant *>(ApplyConversion(expr, route, routelen));
return ex->IntVal;
}
}
double ZCCCompiler::GetDouble(ZCC_Expression *expr)
{
if (expr->Type == TypeError)
{
return 0;
}
const PType::Conversion *route[CONVERSION_ROUTE_SIZE];
int routelen = expr->Type->FindConversion(TypeFloat64, route, countof(route));
if (routelen < 0)
{
Error(expr, "Cannot convert to float");
return 0;
}
else
{
auto ex = static_cast<ZCC_ExprConstant *>(ApplyConversion(expr, route, routelen));
return ex->DoubleVal;
}
}
const char *ZCCCompiler::GetString(ZCC_Expression *expr, bool silent)
{
if (expr->Type == TypeError)
{
- another batch of Hexen items converted. - fixed parsing of named damage types and pain chances.
2016-10-18 08:09:02 +00:00
return nullptr;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
else if (expr->Type->IsKindOf(RUNTIME_CLASS(PString)))
{
return static_cast<ZCC_ExprConstant *>(expr)->StringVal->GetChars();
}
else if (expr->Type->IsKindOf(RUNTIME_CLASS(PName)))
{
// Ugh... What a mess...
return FName(ENamedName(static_cast<ZCC_ExprConstant *>(expr)->IntVal)).GetChars();
}
else
{
if (!silent) Error(expr, "Cannot convert to string");
- another batch of Hexen items converted. - fixed parsing of named damage types and pain chances.
2016-10-18 08:09:02 +00:00
return nullptr;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
}
//==========================================================================
//
// Parses an actor property's parameters and calls the handler
//
//==========================================================================
void ZCCCompiler::DispatchProperty(FPropertyInfo *prop, ZCC_PropertyStmt *property, AActor *defaults, Baggage &bag)
{
static TArray<FPropParam> params;
static TArray<FString> strings;
params.Clear();
strings.Clear();
params.Reserve(1);
params[0].i = 0;
if (prop->params[0] != '0')
{
if (property->Values == nullptr)
{
Error(property, "%s: arguments missing", prop->name);
return;
}
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
property->Values = Simplify(property->Values, &bag.Info->Symbols, true); // need to do this before the loop so that we can find the head node again.
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
const char * p = prop->params;
auto exp = property->Values;
while (true)
{
FPropParam conv;
FPropParam pref;
if (exp->NodeType != AST_ExprConstant)
{
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
// If we get TypeError, there has already been a message from deeper down so do not print another one.
if (exp->Type != TypeError) Error(exp, "%s: non-constant parameter", prop->name);
return;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
conv.s = nullptr;
pref.s = nullptr;
pref.i = -1;
switch ((*p) & 223)
{
case 'X': // Expression in parentheses or number. We only support the constant here. The function will have to be handled by a separate property to get past the parser.
- fixed parsing of 'damage' default property. - fixed processing of goto labels.
2016-10-13 20:52:31 +00:00
conv.i = GetInt(exp);
params.Push(conv);
conv.exp = nullptr;
break;
case 'I':
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
case 'M': // special case for morph styles in DECORATE . This expression-aware parser will not need this.
case 'N': // special case for thing activations in DECORATE. This expression-aware parser will not need this.
conv.i = GetInt(exp);
break;
case 'F':
conv.d = GetDouble(exp);
break;
- another batch of Hexen items converted. - fixed parsing of named damage types and pain chances.
2016-10-18 08:09:02 +00:00
case 'Z': // an optional string. Does not allow any numeric value.
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
if (!GetString(exp, true))
{
// apply this expression to the next argument on the list.
params.Push(conv);
params[0].i++;
p++;
continue;
}
- another batch of Hexen items converted. - fixed parsing of named damage types and pain chances.
2016-10-18 08:09:02 +00:00
conv.s = GetString(exp);
break;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
case 'C': // this parser accepts colors only in string form.
pref.i = 1;
case 'S':
- fixed the octal parser in strbin. Like its hex counterpart it needs to backtrack one character if it find the end of a sequence. - since ZScript already receives filtered strings, the 'T' converter for the properties should not do it again.
2016-11-07 08:54:46 +00:00
case 'T': // a filtered string (ZScript only parses filtered strings so there's nothing to do here.)
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
conv.s = GetString(exp);
break;
case 'L': // Either a number or a list of strings
if (!GetString(exp, true))
{
pref.i = 0;
conv.i = GetInt(exp);
}
else
{
pref.i = 1;
params.Push(pref);
params[0].i++;
do
{
conv.s = GetString(exp);
if (conv.s != nullptr)
{
params.Push(conv);
params[0].i++;
}
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
exp = Simplify(static_cast<ZCC_Expression *>(exp->SiblingNext), &bag.Info->Symbols, true);
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
} while (exp != property->Values);
goto endofparm;
}
break;
default:
assert(false);
break;
}
if (pref.i != -1)
{
params.Push(pref);
params[0].i++;
}
params.Push(conv);
params[0].i++;
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
exp = Simplify(static_cast<ZCC_Expression *>(exp->SiblingNext), &bag.Info->Symbols, true);
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
endofparm:
p++;
// Skip the DECORATE 'no comma' marker
if (*p == '_') p++;
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
if (*p == 0)
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
{
if (exp != property->Values)
{
Error(property, "Too many values for '%s'", prop->name);
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
return;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
break;
}
else if (exp == property->Values)
{
if (*p < 'a')
{
Error(property, "Insufficient parameters for %s", prop->name);
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
return;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
break;
}
}
}
// call the handler
try
{
prop->Handler(defaults, bag.Info, bag, &params[0]);
}
catch (CRecoverableError &error)
{
Error(property, "%s", error.GetMessage());
}
}
//==========================================================================
//
// Parses an actor property
//
//==========================================================================
void ZCCCompiler::ProcessDefaultProperty(PClassActor *cls, ZCC_PropertyStmt *prop, Baggage &bag)
{
auto namenode = prop->Prop;
FString propname;
if (namenode->SiblingNext == namenode)
{
- removed the FxDamageValue hack and implemented it properly using FxReturnStatement. - added handling of damage functions for ZScript.
2016-10-26 12:04:49 +00:00
if (namenode->Id == NAME_DamageFunction)
{
auto x = ConvertNode(prop->Values);
- fixed: DECORATE didn't pass all needed arguments to FunctionBuildList.AddFunction. Removed the default parameter values from this function's declaration to prevent it from happening again.
2016-12-03 17:14:07 +00:00
CreateDamageFunction(cls, (AActor *)bag.Info->Defaults, x, false, Lump);
- fixed: Declaring a damage function must set regular damage to -1 to disable it.
2016-11-07 22:22:00 +00:00
((AActor *)bag.Info->Defaults)->DamageVal = -1;
- removed the FxDamageValue hack and implemented it properly using FxReturnStatement. - added handling of damage functions for ZScript.
2016-10-26 12:04:49 +00:00
return;
}
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
// a one-name property
propname = FName(namenode->Id);
- removed the FxDamageValue hack and implemented it properly using FxReturnStatement. - added handling of damage functions for ZScript.
2016-10-26 12:04:49 +00:00
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
else if (namenode->SiblingNext->SiblingNext == namenode)
{
// a two-name property
propname << FName(namenode->Id) << "." << FName(static_cast<ZCC_Identifier *>(namenode->SiblingNext)->Id);
}
else
{
Error(prop, "Property name may at most contain two parts");
return;
}
- removed the FxDamageValue hack and implemented it properly using FxReturnStatement. - added handling of damage functions for ZScript.
2016-10-26 12:04:49 +00:00
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
FPropertyInfo *property = FindProperty(propname);
if (property != nullptr && property->category != CAT_INFO)
{
if (cls->IsDescendantOf(*property->cls))
{
DispatchProperty(property, prop, (AActor *)bag.Info->Defaults, bag);
}
else
{
Error(prop, "'%s' requires an actor of type '%s'\n", propname.GetChars(), (*property->cls)->TypeName.GetChars());
}
}
else
{
Error(prop, "'%s' is an unknown actor property\n", propname.GetChars());
}
}
//==========================================================================
//
// Finds a flag and sets or clears it
//
//==========================================================================
void ZCCCompiler::ProcessDefaultFlag(PClassActor *cls, ZCC_FlagStmt *flg)
{
auto namenode = flg->name;
const char *n1 = FName(namenode->Id).GetChars(), *n2;
if (namenode->SiblingNext == namenode)
{
// a one-name flag
n2 = nullptr;
}
else if (namenode->SiblingNext->SiblingNext == namenode)
{
// a two-name flag
n2 = FName(static_cast<ZCC_Identifier *>(namenode->SiblingNext)->Id).GetChars();
}
else
{
Error(flg, "Flag name may at most contain two parts");
return;
}
auto fd = FindFlag(cls, n1, n2, true);
if (fd != nullptr)
{
- made the count flags and NOBLOCKMAP and NOSECTOR read-only. These require special treatment to work which can only be done by a setter function.
2016-10-25 07:55:13 +00:00
if (fd->varflags & VARF_Deprecated)
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
{
Warn(flg, "Deprecated flag '%s%s%s' used", n1, n2 ? "." : "", n2 ? n2 : "");
- implement flag variables with the VM's sbit and lbit instructions. - synthesize PField entries from the flag list for AActor. This intentionally excludes the bounce flags for now. - allow deprecated flags that do not call the deprecated flag handler. - disallow constructs like (a = b) = c by not allowing an address request on an assignment operation. - restrict modify/assign on boolean variables to the bit operators. Everything else needs to promote the result to an integer to make sense so it should be disallowed.
2016-10-24 15:18:20 +00:00
}
if (fd->structoffset == -1)
{
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
HandleDeprecatedFlags((AActor*)cls->Defaults, cls, flg->set, fd->flagbit);
}
else
{
ModActorFlag((AActor*)cls->Defaults, fd, flg->set);
}
}
else
{
Error(flg, "Unknown flag '%s%s%s'", n1, n2 ? "." : "", n2 ? n2 : "");
}
}
//==========================================================================
//
// Parses the default list
//
//==========================================================================
void ZCCCompiler::InitDefaults()
{
for (auto c : Classes)
{
// This may be removed if the conditions change, but right now only subclasses of Actor can define a Default block.
if (!c->Type()->IsDescendantOf(RUNTIME_CLASS(AActor)))
{
if (c->Defaults.Size()) Error(c->cls, "%s: Non-actor classes may not have defaults", c->Type()->TypeName.GetChars());
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
if (c->Type()->ParentClass) c->Type()->ParentClass->DeriveData(c->Type());
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
else
{
// This should never happen.
if (c->Type()->Defaults != nullptr)
{
Error(c->cls, "%s already has defaults", c->Type()->TypeName.GetChars());
}
// This can only occur if a native parent is not initialized. In all other cases the sorting of the class list should prevent this from ever happening.
else if (c->Type()->ParentClass->Defaults == nullptr && c->Type() != RUNTIME_CLASS(AActor))
{
Error(c->cls, "Parent class %s of %s is not initialized", c->Type()->ParentClass->TypeName.GetChars(), c->Type()->TypeName.GetChars());
}
else
{
// Copy the parent's defaults and meta data.
auto ti = static_cast<PClassActor *>(c->Type());
TODO: Find a better place to insert the classes in the hierarchy
2016-11-06 10:40:17 +00:00
fixed and completed the special field init code. Strings can now be used as class members, and so can structs which contain strings. - made 'DamageMultiply' an actor property and moved the initialization of ConversationRoot to the property handler for the compiler to get this stuff out of the type classes. - consolidate default initialization into one function which performs all the required setup. The original implementation did this when adding the fields but that cannot work because at that time no defaults have been created yet. - fixed: When deriving a class the child class's defaults also must initialize the copied parent fields with special initialization. This part was completely missing. - removed DECORATE code for parsing native classes because it's no longer needed.
2016-11-11 13:40:32 +00:00
ti->InitializeDefaults();
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
ti->ParentClass->DeriveData(ti);
fixed and completed the special field init code. Strings can now be used as class members, and so can structs which contain strings. - made 'DamageMultiply' an actor property and moved the initialization of ConversationRoot to the property handler for the compiler to get this stuff out of the type classes. - consolidate default initialization into one function which performs all the required setup. The original implementation did this when adding the fields but that cannot work because at that time no defaults have been created yet. - fixed: When deriving a class the child class's defaults also must initialize the copied parent fields with special initialization. This part was completely missing. - removed DECORATE code for parsing native classes because it's no longer needed.
2016-11-11 13:40:32 +00:00
// We need special treatment for this one field in AActor's defaults which cannot be made visible to DECORATE as a property.
// It's better to do this here under controlled conditions than deeper down in the class type classes.
if (ti == RUNTIME_CLASS(AActor))
{
((AActor*)ti->Defaults)->ConversationRoot = 1;
}
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
Baggage bag;
#ifdef _DEBUG
bag.ClassName = c->Type()->TypeName;
#endif
bag.Info = ti;
bag.DropItemSet = false;
bag.StateSet = false;
- fixed checks in Powerup.Type property to properly deal with the differences between DECORATE and ZScript. - properly initialize Baggage everywhere it gets used. - fixed a few items with incorrect Powerup.Type settings that got flagged by the above changes.
2016-11-07 10:53:49 +00:00
bag.fromDecorate = false;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
bag.CurrentState = 0;
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
bag.Lumpnum = c->cls->SourceLump;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
bag.DropItemList = nullptr;
// The actual script position needs to be set per property.
for (auto d : c->Defaults)
{
auto content = d->Content;
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
if (content != nullptr) do
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
{
switch (content->NodeType)
{
case AST_PropertyStmt:
bag.ScriptPosition.FileName = *content->SourceName;
bag.ScriptPosition.ScriptLine = content->SourceLoc;
ProcessDefaultProperty(ti, static_cast<ZCC_PropertyStmt *>(content), bag);
break;
case AST_FlagStmt:
ProcessDefaultFlag(ti, static_cast<ZCC_FlagStmt *>(content));
break;
- Fixed more GCC/Clang warnings.
2016-11-26 11:45:17 +00:00
default:
break;
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
content = static_cast<decltype(content)>(content->SiblingNext);
} while (content != d->Content);
}
- resorted some of thingdef.cpp's contents into more appropriate files. - split FinishActor into several functions. While DECORATE can, ZSCRIPT cannot do all this in one go. - split the state finalization into several class-specific virtual functions.
2016-10-12 18:42:41 +00:00
if (bag.DropItemSet)
{
bag.Info->SetDropItems(bag.DropItemList);
}
- implemented the Defaults property initializer. This uses the same property and flag tables as DECORATE with a few changes: * it sets the parse mode to strict, so that several DECORATE warnings are now errors. * trying to change a deprecated flag will print a warning. * setting of editor numbers, spawn and conversation ID id not possible. Use MAPINFO to do this. * all subclass flags must use the qualified name now (e.g. +ALWAYSPICKUP will print an error.) * the scriptable Damage property is not yet implemented. This will require a special case with a differently named property in the processing function because in the AST it is no longer possible to distinguish between a damage value and a constant damage function.
2016-10-10 22:56:47 +00:00
}
}
}
}
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
void ZCCCompiler::CompileFunction(ZCC_StructWork *c, ZCC_FuncDeclarator *f, bool forclass)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
TArray<PType *> rets(1);
TArray<PType *> args;
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
TArray<uint32_t> argflags;
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
TArray<VMValue> argdefaults;
- more preparations to compile functions: * Allow PFunction to work without a VMFunction being attached. * The Variant for a function must store the prototype itself instead of relying on the VMFunction it points to. Otherwise it would not be possible to reference a prototype during compilation of the function because it does not exist yet. * Give the variant a list of the function's argument's names, because these are also needed to compile the function. * create an anonymous function symbol when the function gets registered to the builder. At this point we have all the needed information to set it up correctly, but later this is no longer the case. This is the most convenient info to have here because it contains everything that's needed to compile the function in the proper context, so it has to be present when starting compilation. * added some preparations to implement special handling for weapons and custom inventory items, which can run action functions in another actor's context. This part is not active yet but the basics are present in SetImplicitArgs.
2016-10-15 12:36:08 +00:00
TArray<FName> argnames;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
rets.Clear();
args.Clear();
argflags.Clear();
bool hasdefault = false;
// For the time being, let's not allow overloading. This may be reconsidered later but really just adds an unnecessary amount of complexity here.
if (AddTreeNode(f->Name, f, &c->TreeNodes, false))
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto t = f->Type;
if (t != nullptr)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
do
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto type = DetermineType(c->Type(), f, f->Name, t, false, false);
if (type->IsKindOf(RUNTIME_CLASS(PStruct)) && type != TypeVector2 && type != TypeVector3)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
// structs and classes only get passed by pointer.
type = NewPointer(type);
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
// TBD: disallow certain types? For now, let everything pass that isn't an array.
rets.Push(type);
t = static_cast<decltype(t)>(t->SiblingNext);
} while (t != f->Type);
}
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
int notallowed = ZCC_Latent | ZCC_Meta | ZCC_ReadOnly | ZCC_FuncConst | ZCC_Abstract;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (f->Flags & notallowed)
{
Error(f, "Invalid qualifiers for %s (%s not allowed)", FName(f->Name).GetChars(), FlagsToString(f->Flags & notallowed).GetChars());
f->Flags &= notallowed;
}
uint32_t varflags = VARF_Method;
int implicitargs = 1;
AFuncDesc *afd = nullptr;
int useflags = SUF_ACTOR | SUF_OVERLAY | SUF_WEAPON | SUF_ITEM;
if (f->UseFlags != nullptr)
{
useflags = 0;
auto p = f->UseFlags;
do
{
switch (p->Id)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
case NAME_Actor:
useflags |= SUF_ACTOR;
break;
case NAME_Overlay:
useflags |= SUF_OVERLAY;
break;
case NAME_Weapon:
useflags |= SUF_WEAPON;
break;
case NAME_Item:
useflags |= SUF_ITEM;
break;
default:
Error(p, "Unknown Action qualifier %s", FName(p->Id).GetChars());
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
- implemented usage restrictions for action functions.
2016-11-15 20:38:12 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
p = static_cast<decltype(p)>(p->SiblingNext);
} while (p != f->UseFlags);
}
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
// map to implementation flags.
if (f->Flags & ZCC_Private) varflags |= VARF_Private;
if (f->Flags & ZCC_Protected) varflags |= VARF_Protected;
if (f->Flags & ZCC_Deprecated) varflags |= VARF_Deprecated;
if (f->Flags & ZCC_Virtual) varflags |= VARF_Virtual;
if (f->Flags & ZCC_Override) varflags |= VARF_Override;
if (f->Flags & ZCC_Action)
{
// Non-Actors cannot have action functions.
if (!c->Type()->IsKindOf(RUNTIME_CLASS(PClassActor)))
{
Error(f, "'Action' can only be used in child classes of Actor");
}
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
varflags |= VARF_Final; // Action implies Final.
if (useflags & (SUF_OVERLAY | SUF_WEAPON | SUF_ITEM))
{
varflags |= VARF_Action;
implicitargs = 3;
}
else
{
implicitargs = 1;
}
}
if (f->Flags & ZCC_Static) varflags = (varflags & ~VARF_Method) | VARF_Final, implicitargs = 0; // Static implies Final.
- implemented usage restrictions for action functions.
2016-11-15 20:38:12 +00:00
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (varflags & VARF_Override) varflags &= ~VARF_Virtual; // allow 'virtual override'.
// Only one of these flags may be used.
static int exclude[] = { ZCC_Virtual, ZCC_Override, ZCC_Action, ZCC_Static };
static const char * print[] = { "virtual", "override", "action", "static" };
int fc = 0;
FString build;
for (int i = 0; i < 4; i++)
{
if (f->Flags & exclude[i])
{
fc++;
if (build.Len() > 0) build += ", ";
build += print[i];
}
}
if (fc > 1)
{
Error(f, "Invalid combination of qualifiers %s on function %s.", FName(f->Name).GetChars(), build.GetChars());
varflags |= VARF_Method;
}
if (varflags & VARF_Override) varflags |= VARF_Virtual; // Now that the flags are checked, make all override functions virtual as well.
if (f->Flags & ZCC_Native)
{
varflags |= VARF_Native;
afd = FindFunction(c->Type(), FName(f->Name).GetChars());
if (afd == nullptr)
{
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
Error(f, "The function '%s.%s' has not been exported from the executable.", c->Type()->TypeName.GetChars(), FName(f->Name).GetChars());
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
}
else
{
(*afd->VMPointer)->ImplicitArgs = BYTE(implicitargs);
}
}
SetImplicitArgs(&args, &argflags, &argnames, c->Type(), varflags, useflags);
argdefaults.Resize(argnames.Size());
auto p = f->Params;
bool hasoptionals = false;
if (p != nullptr)
{
do
{
int elementcount = 1;
VMValue vmval[3]; // default is REGT_NIL which means 'no default value' here.
if (p->Type != nullptr)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto type = DetermineType(c->Type(), p, f->Name, p->Type, false, false);
int flags = 0;
- scriptified a_strifeitems.cpp and a_debris.cpp. - Changed the glass shards so that they do not have to override FloorBounceMissile. It was the only place where this was virtually overridden and provided little usefulness. - made 'out' variables work. - fixed virtual call handling for HandlePickup.
2016-11-29 11:17:05 +00:00
if (type->IsA(RUNTIME_CLASS(PStruct)) && type != TypeVector2 && type != TypeVector3)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- scriptified a_strifeitems.cpp and a_debris.cpp. - Changed the glass shards so that they do not have to override FloorBounceMissile. It was the only place where this was virtually overridden and provided little usefulness. - made 'out' variables work. - fixed virtual call handling for HandlePickup.
2016-11-29 11:17:05 +00:00
// Structs are being passed by pointer, but unless marked 'out' that pointer must be readonly.
type = NewPointer(type /*, !(p->Flags & ZCC_Out)*/);
flags |= VARF_Ref;
}
else if (type->GetRegType() != REGT_NIL)
{
if (p->Flags & ZCC_Out) flags |= VARF_Out;
if (type == TypeVector2)
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
{
elementcount = 2;
}
else if (type == TypeVector3)
{
elementcount = 3;
}
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (type->GetRegType() == REGT_NIL && type != TypeVector2 && type != TypeVector3)
- scriptified A_BruisAttack. - removed 'self' as a dedicated token. Internally this gets handled as a normal but implicitly named variable so the token just gets in the way of proper processing. - removed P_ prefix from SpawnMissile export. - fixed a crash with misnamed function exports.
2016-10-23 15:15:24 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
Error(p, "Invalid type %s for function parameter", type->DescriptiveName());
- scriptified A_BruisAttack. - removed 'self' as a dedicated token. Internally this gets handled as a normal but implicitly named variable so the token just gets in the way of proper processing. - removed P_ prefix from SpawnMissile export. - fixed a crash with misnamed function exports.
2016-10-23 15:15:24 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else if (p->Default != nullptr)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
flags |= VARF_Optional;
hasoptionals = true;
// The simplifier is not suited to convert the constant into something usable.
// All it does is reduce the expression to a constant but we still got to do proper type checking and conversion.
// It will also lose important type info about enums, once these get implemented
// The code generator can do this properly for us.
FxExpression *x = new FxTypeCast(ConvertNode(p->Default), type, false);
FCompileContext ctx(c->Type(), false);
x = x->Resolve(ctx);
if (x != nullptr)
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
// Vectors need special treatment because they use more than one entry in the Defaults and do not report as actual constants
if (type == TypeVector2 && x->ExprType == EFX_VectorValue && static_cast<FxVectorValue *>(x)->isConstVector(2))
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto vx = static_cast<FxVectorValue *>(x);
vmval[0] = static_cast<FxConstant *>(vx->xyz[0])->GetValue().GetFloat();
vmval[1] = static_cast<FxConstant *>(vx->xyz[1])->GetValue().GetFloat();
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else if (type == TypeVector3 && x->ExprType == EFX_VectorValue && static_cast<FxVectorValue *>(x)->isConstVector(3))
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto vx = static_cast<FxVectorValue *>(x);
vmval[0] = static_cast<FxConstant *>(vx->xyz[0])->GetValue().GetFloat();
vmval[1] = static_cast<FxConstant *>(vx->xyz[1])->GetValue().GetFloat();
vmval[2] = static_cast<FxConstant *>(vx->xyz[2])->GetValue().GetFloat();
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else if (!x->isConstant())
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
Error(p, "Default parameter %s is not constant in %s", FName(p->Name).GetChars(), FName(f->Name).GetChars());
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else if (x->ValueType != type)
- implemented passing vectors as parameters. So far working for native functions. - removed the bogus optional value from the first A_Jump argument. A quick test with an older ZDoom revealed that this was never working - and implementing it would make things a lot more complicated, especially error checking in the code generator. - fixed: The check for insufficient parameters to a function call was missing.
2016-10-29 11:10:27 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
Error(p, "Default parameter %s could not be converted to target type %s", FName(p->Name).GetChars(), c->Type()->TypeName.GetChars());
- implemented passing vectors as parameters. So far working for native functions. - removed the bogus optional value from the first A_Jump argument. A quick test with an older ZDoom revealed that this was never working - and implementing it would make things a lot more complicated, especially error checking in the code generator. - fixed: The check for insufficient parameters to a function call was missing.
2016-10-29 11:10:27 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else
{
auto cnst = static_cast<FxConstant *>(x);
hasdefault = true;
switch (type->GetRegType())
{
case REGT_INT:
vmval[0] = cnst->GetValue().GetInt();
break;
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
case REGT_FLOAT:
vmval[0] = cnst->GetValue().GetFloat();
break;
case REGT_POINTER:
if (type->IsKindOf(RUNTIME_CLASS(PClassPointer)))
vmval[0] = (DObject*)cnst->GetValue().GetPointer();
else
vmval[0] = cnst->GetValue().GetPointer();
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
case REGT_STRING:
vmval[0] = cnst->GetValue().GetString();
break;
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
default:
assert(0 && "no valid type for constant");
break;
}
}
}
if (x != nullptr) delete x;
}
else if (hasoptionals)
- named class functions are working. Yay!! - converted A_BspiAttack and A_BabyMetal to script functions to test the implementation.
2016-10-23 12:26:33 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
Error(p, "All arguments after the first optional one need also be optional.");
- named class functions are working. Yay!! - converted A_BspiAttack and A_BabyMetal to script functions to test the implementation.
2016-10-23 12:26:33 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
// TBD: disallow certain types? For now, let everything pass that isn't an array.
args.Push(type);
argflags.Push(flags);
argnames.Push(p->Name);
- named class functions are working. Yay!! - converted A_BspiAttack and A_BabyMetal to script functions to test the implementation.
2016-10-23 12:26:33 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
args.Push(nullptr);
argflags.Push(0);
argnames.Push(NAME_None);
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
for (int i = 0; i<elementcount; i++) argdefaults.Push(vmval[i]);
p = static_cast<decltype(p)>(p->SiblingNext);
} while (p != f->Params);
}
PFunction *sym = new PFunction(c->Type(), f->Name);
sym->AddVariant(NewPrototype(rets, args), argflags, argnames, afd == nullptr ? nullptr : *(afd->VMPointer), varflags, useflags);
c->Type()->Symbols.ReplaceSymbol(sym);
- scriptified the Heresiarch.
2016-11-29 16:17:10 +00:00
auto cls = dyn_cast<PClass>(c->Type());
PFunction *virtsym = nullptr;
if (cls != nullptr && cls->ParentClass != nullptr) virtsym = dyn_cast<PFunction>(cls->ParentClass->Symbols.FindSymbol(FName(f->Name), true));
unsigned vindex = ~0u;
if (virtsym != nullptr) vindex = virtsym->Variants[0].Implementation->VirtualIndex;
if (vindex != ~0u || (varflags & VARF_Virtual))
{
// Todo: Check if the declaration is legal.
// First step: compare prototypes - if they do not match the virtual base method does not apply.
// Second step: Check flags. Possible cases:
// 1. Base method is final: Error.
// 2. This method is override: Base virtual method must exist
// 3. This method is virtual but not override: Base may not have a virtual method with the same prototype.
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (!(f->Flags & ZCC_Native))
{
- scriptified A_Saw. - implemented multiple-return-value assignment. Due to some grammar conflicts the originally intended Lua-inspired syntax of 'a, b = Function()' could not be done, so it's '[a, b] = Function()'
2016-11-19 23:25:38 +00:00
if (f->Body == nullptr)
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
{
- scriptified A_Saw. - implemented multiple-return-value assignment. Due to some grammar conflicts the originally intended Lua-inspired syntax of 'a, b = Function()' could not be done, so it's '[a, b] = Function()'
2016-11-19 23:25:38 +00:00
Error(f, "Empty function %s", FName(f->Name).GetChars());
return;
}
else
{
auto code = ConvertAST(c->Type(), f->Body);
if (code != nullptr)
{
- added per-channel access for color variables. However, since they are locally stored in registers, it required a minor bit of fudging for this case. - make sure that a PFunction's implementation is always initialized before starting the code generator.
2016-11-21 11:38:39 +00:00
FunctionBuildList.AddFunction(sym, code, FStringf("%s.%s", c->Type()->TypeName.GetChars(), FName(f->Name).GetChars()), false, -1, 0, Lump);
- scriptified A_Saw. - implemented multiple-return-value assignment. Due to some grammar conflicts the originally intended Lua-inspired syntax of 'a, b = Function()' could not be done, so it's '[a, b] = Function()'
2016-11-19 23:25:38 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
}
}
if (sym->Variants[0].Implementation != nullptr && hasdefault) // do not copy empty default lists, they only waste space and processing time.
{
sym->Variants[0].Implementation->DefaultArgs = std::move(argdefaults);
}
if (varflags & VARF_Virtual)
{
- scriptified the Ironlich.
2016-11-23 20:26:59 +00:00
if (sym->Variants[0].Implementation == nullptr)
{
Error(f, "Virtual function %s.%s not present.", c->Type()->TypeName.GetChars(), FName(f->Name).GetChars());
return;
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (varflags & VARF_Final)
{
sym->Variants[0].Implementation->Final = true;
}
if (forclass)
{
PClass *clstype = static_cast<PClass *>(c->Type());
int vindex = clstype->FindVirtualIndex(sym->SymbolName, sym->Variants[0].Proto);
// specifying 'override' is necessary to prevent one of the biggest problem spots with virtual inheritance: Mismatching argument types.
if (varflags & VARF_Override)
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
if (vindex == -1)
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
{
- scriptified the scripted marines. - fixed symbol name generation for native functions. - moved PrintableName to VMFunction so that native functions also have this information.
2016-11-21 18:09:58 +00:00
Error(f, "Attempt to override non-existent virtual function %s", FName(f->Name).GetChars());
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
{
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto oldfunc = clstype->Virtuals[vindex];
if (oldfunc->Final)
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
{
- scriptified the scripted marines. - fixed symbol name generation for native functions. - moved PrintableName to VMFunction so that native functions also have this information.
2016-11-21 18:09:58 +00:00
Error(f, "Attempt to override final function %s", FName(f->Name).GetChars());
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
clstype->Virtuals[vindex] = sym->Variants[0].Implementation;
sym->Variants[0].Implementation->VirtualIndex = vindex;
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
}
else
{
if (vindex != -1)
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
{
- allow the VM to run on one global stack per thread. It is utterly pointless to require every function that wants to make a VM call to allocate a new stack first. The allocation overhead doubles the time to set up the call. With one stack, previously allocated memory can be reused. The only important thing is, if this ever gets used in a multithreaded environment to have the stack being declared as thread_local, although for ZDoom this is of no consequence. - eliminated all cases where native code was calling other native code through the VM interface. After scriptifying the game code, only 5 places were left which were quickly eliminated. This was mostly to ensure that the native VM function parameters do not need to be propagated further than absolutely necessary.
2016-11-30 16:15:01 +00:00
Error(f, "Function %s attempts to override parent function without 'override' qualifier", FName(f->Name).GetChars());
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
sym->Variants[0].Implementation->VirtualIndex = clstype->Virtuals.Push(sym->Variants[0].Implementation);
- implemented handling for virtual function. Syntax-wise I chose to make it as strict as possible to reduce the chance of errors: Virtual base functions must be declared with the 'virtual' keyword, and overrides in child classes with the 'override' keyword. This way any mismatch in parameters that otherwise would cause silent failure will outright produce a compile error.
2016-11-11 19:05:07 +00:00
}
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
else
{
Error(p, "Virtual functions can only be defined for classes");
}
}
}
}
//==========================================================================
//
// Parses the functions list
//
//==========================================================================
void ZCCCompiler::InitFunctions()
{
for (auto s : Structs)
{
for (auto f : s->Functions)
{
CompileFunction(s, f, false);
}
}
for (auto c : Classes)
{
// cannot be done earlier because it requires the parent class to be processed by this code, too.
if (c->Type()->ParentClass != nullptr)
{
c->Type()->Virtuals = c->Type()->ParentClass->Virtuals;
}
for (auto f : c->Functions)
{
CompileFunction(c, f, true);
- initialize function symbols. At the moment all it does is process the existing native functions. Any further processing will be done once the base classes of the engine can be parsed from the scripting files. - switched the types of the internal 'self' and 'stateowner' parameters so that they get assigned correctly. I can't tell if this will error out if fields get accessed from the caller with the wrong class, but for actual scripting to work these must be correct. The committed 'actor.txt' can be parsed successfully, with the exception of a few subclass references that cannot be resolved yet.
2016-10-11 16:53:10 +00:00
}
}
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
//==========================================================================
//
// very complicated check for random duration.
//
//==========================================================================
static bool CheckRandom(ZCC_Expression *duration)
{
if (duration->NodeType != AST_ExprFuncCall) return false;
auto func = static_cast<ZCC_ExprFuncCall *>(duration);
if (func->Function == nullptr) return false;
if (func->Function->NodeType != AST_ExprID) return false;
auto f2 = static_cast<ZCC_ExprID *>(func->Function);
return f2->Identifier == NAME_Random;
}
//==========================================================================
//
// Sets up the action function call
//
//==========================================================================
- scriptified the Sigil. This isn't fully tested yet.
2016-11-28 13:39:25 +00:00
FxExpression *ZCCCompiler::SetupActionFunction(PClass *cls, ZCC_TreeNode *af, int StateFlags)
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
{
// We have 3 cases to consider here:
- scriptified A_Mushroom to test something a bit more complex. - fixed: FxMinusSign trashed local variables that were used with negation. - fixed: FxConditional only handled ints and floats, but not pointers and strings. - fixed: A 'no states in non-actors' error was triggered, even for classes without any states.
2016-10-30 13:00:11 +00:00
// 1. A function without parameters. This can be called directly
// 2. A functon with parameters. This needs to be wrapped into a helper function to set everything up.
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
// 3. An anonymous function.
// 1. and 2. are exposed through AST_ExprFunctionCall
if (af->NodeType == AST_ExprFuncCall)
{
auto fc = static_cast<ZCC_ExprFuncCall *>(af);
assert(fc->Function->NodeType == AST_ExprID);
auto id = static_cast<ZCC_ExprID *>(fc->Function);
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
// We must skip ACS_NamedExecuteWithResult here, because this name both exists as an action function and as a builtin.
// The code which gets called from here can easily make use of the builtin, so let's just pass this name without any checks.
// The actual action function is still needed by DECORATE:
if (id->Identifier != NAME_ACS_NamedExecuteWithResult)
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
{
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
PFunction *afd = dyn_cast<PFunction>(cls->Symbols.FindSymbol(id->Identifier, true));
if (afd != nullptr)
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
{
- allow calling non-action functions directly from a state. Ironically this only requires a very minor change in the calling code and an added member for the VMFunction to tell that code how many parameters to pass. This change will allow to turn the vast majority of action functions into regular members, the only ones that still need to be an action function are the few that actually use the pointers.
2016-10-22 14:35:48 +00:00
if (fc->Parameters == nullptr && !(afd->Variants[0].Flags & VARF_Virtual))
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
{
- Fixed Clang errors with delayed templates on.
2016-11-11 14:22:53 +00:00
FArgumentList argumentlist;
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
// We can use this function directly without wrapping it in a caller.
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
auto selfclass = dyn_cast<PClass>(afd->Variants[0].SelfClass);
assert(selfclass != nullptr); // non classes are not supposed to get here.
- scriptified the Sigil. This isn't fully tested yet.
2016-11-28 13:39:25 +00:00
int comboflags = afd->Variants[0].UseFlags & StateFlags;
if (comboflags == StateFlags) // the function must satisfy all the flags the state requires
- fixed: All functions that are callable from weapon states and not members of Actor need to be declared 'action'. With the stricter type checks of the self pointer that were now implemented these all produced errors.
2016-11-13 13:20:30 +00:00
{
return new FxVMFunctionCall(new FxSelf(*af), afd, argumentlist, *af, false);
}
else
{
Error(af, "Cannot use non-action function %s here.", FName(id->Identifier).GetChars());
}
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
else
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
{
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
// it may also be an action special so check that first before printing an error.
if (!P_FindLineSpecial(FName(id->Identifier).GetChars()))
{
Error(af, "%s: action function not found in %s", FName(id->Identifier).GetChars(), cls->TypeName.GetChars());
return nullptr;
}
// Action specials fall through to the code generator.
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
}
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
return ConvertAST(cls, af);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
//==========================================================================
//
// Compile the states
//
//==========================================================================
void ZCCCompiler::CompileStates()
{
for (auto c : Classes)
{
- scriptified A_Mushroom to test something a bit more complex. - fixed: FxMinusSign trashed local variables that were used with negation. - fixed: FxConditional only handled ints and floats, but not pointers and strings. - fixed: A 'no states in non-actors' error was triggered, even for classes without any states.
2016-10-30 13:00:11 +00:00
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (!c->Type()->IsDescendantOf(RUNTIME_CLASS(AActor)))
{
- scriptified A_Mushroom to test something a bit more complex. - fixed: FxMinusSign trashed local variables that were used with negation. - fixed: FxConditional only handled ints and floats, but not pointers and strings. - fixed: A 'no states in non-actors' error was triggered, even for classes without any states.
2016-10-30 13:00:11 +00:00
if (c->States.Size()) Error(c->cls, "%s: States can only be defined for actors.", c->Type()->TypeName.GetChars());
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
continue;
}
TODO: Find a better place to insert the classes in the hierarchy
2016-11-06 10:40:17 +00:00
// Same here, hack in the DVMObject as they weren't in the list originally
// TODO: process them in a non hackish way obviously
if (c->Type()->bRuntimeClass == true && c->Type()->ParentClass->bRuntimeClass == false)
{
auto vmtype = static_cast<PClassActor *>(c->Type()->ParentClass);
if (vmtype->StateList == nullptr)
{
FStateDefinitions vmstates;
vmstates.MakeStateDefines(dyn_cast<PClassActor>(vmtype->ParentClass));
vmtype->Finalize(vmstates);
}
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
FString statename; // The state builder wants the label as one complete string, not separated into tokens.
FStateDefinitions statedef;
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
statedef.MakeStateDefines(dyn_cast<PClassActor>(c->Type()->ParentClass));
- fixed: The states parser tried to simplify the head node for random duration but it has to simplify the two value nodes separately. - added jump by index to ZScript.
2016-11-04 23:39:00 +00:00
int numframes = 0;
- removed token 'mode' because it isn't used anywhere and clashed with some actor properties. - fixed uninitialized counter variable in DECORATE parser. - allow dottable_id of xxx.color so that the property parser can parse 'powerup.color'. - fixed crash with actor replacement in script compiler. - add the lump number to tree nodes because parts of the property parser need that to make decisions. - removed test stuff. - converted inventory.txt, player.txt and specialspot.txt to ZSCRIPT. These were the minimal files required to allow actor.txt to parse successfully. - removed the converted files from the DECORATE include list so that these are entirely handled by ZSCRIPT now.
2016-10-13 18:45:52 +00:00
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
for (auto s : c->States)
{
- finished the state usage parser. - added state usage specifiers to Actor and Inventory. The states in these classes must be set to full access so that any existing mod can link to them.
2016-11-14 22:24:10 +00:00
int flags;
if (s->Flags != nullptr)
{
flags = 0;
auto p = s->Flags;
do
{
switch (p->Id)
{
case NAME_Actor:
flags |= SUF_ACTOR;
break;
case NAME_Overlay:
flags |= SUF_OVERLAY;
break;
case NAME_Weapon:
flags |= SUF_WEAPON;
break;
case NAME_Item:
flags |= SUF_ITEM;
break;
default:
Error(p, "Unknown States qualifier %s", FName(p->Id).GetChars());
break;
}
p = static_cast<decltype(p)>(p->SiblingNext);
} while (p != s->Flags);
}
else
{
flags = static_cast<PClassActor *>(c->Type())->DefaultStateUsage;
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
auto st = s->Body;
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
if (st != nullptr) do
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
{
switch (st->NodeType)
{
case AST_StateLabel:
{
auto sl = static_cast<ZCC_StateLabel *>(st);
statename = FName(sl->Label);
statedef.AddStateLabel(statename);
break;
}
case AST_StateLine:
{
auto sl = static_cast<ZCC_StateLine *>(st);
FState state;
memset(&state, 0, sizeof(state));
- added runtime checks for the state usage flags.
2016-11-14 23:18:57 +00:00
state.UseFlags = flags;
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (sl->Sprite->Len() != 4)
{
Error(sl, "Sprite name must be exactly 4 characters. Found '%s'", sl->Sprite->GetChars());
}
else
{
state.sprite = GetSpriteIndex(sl->Sprite->GetChars());
}
// It is important to call CheckRandom before Simplify, because Simplify will resolve the function's name to nonsense
if (CheckRandom(sl->Duration))
{
- fixed: The states parser tried to simplify the head node for random duration but it has to simplify the two value nodes separately. - added jump by index to ZScript.
2016-11-04 23:39:00 +00:00
auto func = static_cast<ZCC_ExprFuncCall *>(sl->Duration);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (func->Parameters == func->Parameters->SiblingNext || func->Parameters != func->Parameters->SiblingNext->SiblingNext)
{
Error(sl, "Random duration requires exactly 2 parameters");
}
- fixed: The states parser tried to simplify the head node for random duration but it has to simplify the two value nodes separately. - added jump by index to ZScript.
2016-11-04 23:39:00 +00:00
auto p1 = Simplify(func->Parameters->Value, &c->Type()->Symbols, true);
auto p2 = Simplify(static_cast<ZCC_FuncParm *>(func->Parameters->SiblingNext)->Value, &c->Type()->Symbols, true);
int v1 = GetInt(p1);
int v2 = GetInt(p2);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (v1 > v2) std::swap(v1, v2);
state.Tics = (int16_t)clamp<int>(v1, 0, INT16_MAX);
state.TicRange = (uint16_t)clamp<int>(v2 - v1, 0, UINT16_MAX);
}
else
{
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
auto duration = Simplify(sl->Duration, &c->Type()->Symbols, true);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (duration->Operation == PEX_ConstValue)
{
state.Tics = (int16_t)clamp<int>(GetInt(duration), -1, INT16_MAX);
state.TicRange = 0;
}
else
{
Error(sl, "Duration is not a constant");
}
}
- implemented State as an actual native struct, so that its fields can be accessed from scripts. - refactored state bitfield members into a flag word because the address of a bitfield cannot be taken, making such variables inaccessible to scripts. - actually use PNativeStruct for representing native structs defined in a script.
2016-11-21 20:34:34 +00:00
if (sl->bBright) state.StateFlags |= STF_FULLBRIGHT;
if (sl->bFast) state.StateFlags |= STF_FAST;
if (sl->bSlow) state.StateFlags |= STF_SLOW;
if (sl->bCanRaise) state.StateFlags |= STF_CANRAISE;
if (sl->bNoDelay) state.StateFlags |= STF_NODELAY;
if (sl->bNoDelay)
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
{
if (statedef.GetStateLabelIndex(NAME_Spawn) != statedef.GetStateCount())
{
Warn(sl, "NODELAY only has an effect on the first state after 'Spawn:'");
}
}
if (sl->Offset != nullptr)
{
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
auto o1 = static_cast<ZCC_Expression *>(Simplify(sl->Offset, &c->Type()->Symbols, true));
auto o2 = static_cast<ZCC_Expression *>(Simplify(static_cast<ZCC_Expression *>(o1->SiblingNext), &c->Type()->Symbols, true));
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (o1->Operation != PEX_ConstValue || o2->Operation != PEX_ConstValue)
{
Error(o1, "State offsets must be constant");
}
else
{
state.Misc1 = GetInt(o1);
state.Misc2 = GetInt(o2);
}
}
#ifdef DYNLIGHT
if (sl->Lights != nullptr)
{
auto l = sl->Lights;
do
{
AddStateLight(&state, GetString(l));
l = static_cast<decltype(l)>(l->SiblingNext);
} while (l != sl->Lights);
}
#endif
if (sl->Action != nullptr)
{
- scriptified the Sigil. This isn't fully tested yet.
2016-11-28 13:39:25 +00:00
auto code = SetupActionFunction(static_cast<PClassActor *>(c->Type()), sl->Action, state.UseFlags);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (code != nullptr)
{
- added compile time checks for bad state links and state label references. - preserve a state's source line information for the postprocessing phase so that the checker can output more useful information. - added missing check for weapon psprites to DPSprite::SetState.
2016-11-15 10:21:08 +00:00
auto funcsym = CreateAnonymousFunction(c->Type(), nullptr, state.UseFlags);
- added a new variable flag that allows defining variables which can only be modified by internal script code but not by external mods. Currently this is used by the NoBlockmap and NoSector flags which need special handling for changing.
2016-11-08 10:12:56 +00:00
state.ActionFunc = FunctionBuildList.AddFunction(funcsym, code, FStringf("%s.StateFunction.%d", c->Type()->TypeName.GetChars(), statedef.GetStateCount()), false, statedef.GetStateCount(), (int)sl->Frames->Len(), Lump);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
}
- fixed order of state setup. It looks like the setup of the action function and the adding of the states got inverted by something I tried, leaving the ActionFunc member empty on the real states.
2016-10-14 19:56:45 +00:00
- added compile time checks for bad state links and state label references. - preserve a state's source line information for the postprocessing phase so that the checker can output more useful information. - added missing check for weapon psprites to DPSprite::SetState.
2016-11-15 10:21:08 +00:00
int count = statedef.AddStates(&state, sl->Frames->GetChars(), *sl);
- fixed order of state setup. It looks like the setup of the action function and the adding of the states got inverted by something I tried, leaving the ActionFunc member empty on the real states.
2016-10-14 19:56:45 +00:00
if (count < 0)
{
Error(sl, "Invalid frame character string '%s'", sl->Frames->GetChars());
count = -count;
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
break;
}
case AST_StateGoto:
{
auto sg = static_cast<ZCC_StateGoto *>(st);
statename = "";
if (sg->Qualifier != nullptr)
{
- converted all of Doom's actors. - fixed a few problems that were encountered during conversion: * action specials as action functions were not recognized by the parser. * Player.StartItem could not be parsed. * disabled the naming hack for PowerupType. ZScript, unlike DECORATE will never prepend 'Power' to the power's name, it always needs to specified by its full name. * states and defaults were not checked for empty bodies. * the scope qualifier for goto labels was not properly converted to a string, because it is an ENamedName, not an FName.
2016-10-14 18:08:41 +00:00
statename << FName(sg->Qualifier->Id) << "::";
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
auto part = sg->Label;
do
{
- fixed parsing of 'damage' default property. - fixed processing of goto labels.
2016-10-13 20:52:31 +00:00
statename << FName(part->Id) << '.';
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
part = static_cast<decltype(part)>(part->SiblingNext);
} while (part != sg->Label);
statename.Truncate((long)statename.Len() - 1); // remove the last '.' in the label name
if (sg->Offset != nullptr)
{
- Changed ZCCCompiler::Simplify to allow running in constant and non-constant mode. The difference is that non-constant mode does not attempt to simplify function name expressions because these cannot be processed without destroying them, if some identifier in there is not referencing a symbol, which is quite common.
2016-10-15 23:33:36 +00:00
auto ofs = Simplify(sg->Offset, &c->Type()->Symbols, true);
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
if (ofs->Operation != PEX_ConstValue)
{
Error(sg, "Constant offset expected for GOTO");
}
else
{
int offset = GetInt(ofs);
if (offset < 0)
{
Error(sg, "GOTO offset must be positive");
offset = 0;
}
if (offset > 0)
{
statename.AppendFormat("+%d", offset);
}
}
}
if (!statedef.SetGotoLabel(statename))
{
Error(sg, "GOTO before first state");
}
break;
}
case AST_StateFail:
case AST_StateWait:
if (!statedef.SetWait())
{
Error(st, "%s before first state", st->NodeType == AST_StateFail ? "Fail" : "Wait");
continue;
}
break;
case AST_StateLoop:
if (!statedef.SetLoop())
{
Error(st, "LOOP before first state");
continue;
}
break;
case AST_StateStop:
if (!statedef.SetStop())
{
Error(st, "STOP before first state");
}
break;
default:
assert(0 && "Bad AST node in state");
}
st = static_cast<decltype(st)>(st->SiblingNext);
} while (st != s->Body);
}
- resorted some of thingdef.cpp's contents into more appropriate files. - split FinishActor into several functions. While DECORATE can, ZSCRIPT cannot do all this in one go. - split the state finalization into several class-specific virtual functions.
2016-10-12 18:42:41 +00:00
try
{
static_cast<PClassActor *>(c->Type())->Finalize(statedef);
}
catch (CRecoverableError &err)
{
Error(c->cls, "%s", err.GetMessage());
}
- implemented the state compiler. So far all it can handle is parameter-less functions. To do the rest, some cleanup is needed first, to untangle the DECORATE parser from the actual code generation so that the low end stuff can actually be reused here instead of having to be redone.
2016-10-12 15:50:23 +00:00
}
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
//==========================================================================
//
// Convert the AST data for the code generator.
//
//==========================================================================
- allow functions in structs. This is needed for several internal data types like players, sectors, lines, etc. - added a new type 'NativeStruct'. This will be used for types that cannot be instantiated, and is also needed to cleanly handle many internal types that only can exist as reference.
2016-11-19 11:12:29 +00:00
FxExpression *ZCCCompiler::ConvertAST(PStruct *cls, ZCC_TreeNode *ast)
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
ConvertClass = cls;
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
// there are two possibilities here: either a single function call or a compound statement. For a compound statement we also need to check if the last thing added was a return.
if (ast->NodeType == AST_ExprFuncCall)
{
- fixed identifier resolution in FxIdentifier for action functions with a different invoker than self. - fixed creation of direct function invocations on a state line. In order to receive the implicit arguments this needs to be wrapped into a compound statement so that the local variable getter works.
2016-11-12 14:21:59 +00:00
auto cp = new FxCompoundStatement(*ast);
cp->Add(new FxReturnStatement(ConvertNode(ast), *ast));
return cp;
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
}
else
{
// This must be done here so that we can check for a trailing return statement.
auto x = new FxCompoundStatement(*ast);
auto compound = static_cast<ZCC_CompoundStmt *>(ast);
- fixed: Tacking on a return statement should only be done if the function has branches that actually reach the end. Otherwise it may interfere with return type deduction. - used the return check to optimize out unneeded jumps at the end of an if statement's first block.
2016-11-05 17:05:57 +00:00
//bool isreturn = false;
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
auto node = compound->Content;
if (node != nullptr) do
{
x->Add(ConvertNode(node));
- fixed: Tacking on a return statement should only be done if the function has branches that actually reach the end. Otherwise it may interfere with return type deduction. - used the return check to optimize out unneeded jumps at the end of an if statement's first block.
2016-11-05 17:05:57 +00:00
//isreturn = node->NodeType == AST_ReturnStmt;
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
node = static_cast<decltype(node)>(node->SiblingNext);
} while (node != compound->Content);
- fixed: Tacking on a return statement should only be done if the function has branches that actually reach the end. Otherwise it may interfere with return type deduction. - used the return check to optimize out unneeded jumps at the end of an if statement's first block.
2016-11-05 17:05:57 +00:00
//if (!isreturn) x->Add(new FxReturnStatement(nullptr, *ast));
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
return x;
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
}
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
#define xx(a,z) z,
static int Pex2Tok[] = {
#include "zcc_exprlist.h"
};
//==========================================================================
//
// Helper for modify/assign operators
//
//==========================================================================
static FxExpression *ModifyAssign(FxBinary *operation, FxExpression *left)
{
auto assignself = static_cast<FxAssignSelf *>(operation->left);
- implement flag variables with the VM's sbit and lbit instructions. - synthesize PField entries from the flag list for AActor. This intentionally excludes the bounce flags for now. - allow deprecated flags that do not call the deprecated flag handler. - disallow constructs like (a = b) = c by not allowing an address request on an assignment operation. - restrict modify/assign on boolean variables to the bit operators. Everything else needs to promote the result to an integer to make sense so it should be disallowed.
2016-10-24 15:18:20 +00:00
auto assignment = new FxAssign(left, operation, true);
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
assignself->Assignment = assignment;
return assignment;
}
//==========================================================================
//
// Convert an AST node and its children
//
//==========================================================================
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
FxExpression *ZCCCompiler::ConvertNode(ZCC_TreeNode *ast)
{
- added null pointer. - fixed: FxMemberIdentifier checked for ClassPointers instead of object pointers to resolve the left hand side of the expression. - allow comparison of pointers.
2016-10-21 22:50:04 +00:00
if (ast == nullptr) return nullptr;
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
// Note: Do not call 'Simplify' here because that function tends to destroy identifiers due to lack of context in which to resolve them.
// The Fx nodes created here will be better suited for that.
switch (ast->NodeType)
{
case AST_ExprFuncCall:
{
auto fcall = static_cast<ZCC_ExprFuncCall *>(ast);
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
// function names can either be
// - plain identifiers
// - class members
// - array syntax for random() calls.
// Everything else coming here is a syntax error.
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
FArgumentList args;
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
switch (fcall->Function->NodeType)
{
case AST_ExprID:
// The function name is a simple identifier.
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
return new FxFunctionCall(static_cast<ZCC_ExprID *>(fcall->Function)->Identifier, NAME_None, ConvertNodeList(args, fcall->Parameters), *ast);
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
case AST_ExprMemberAccess:
- fixed a conversion warning with the pointer-type AActor fields. - added master and tracer to the list of exported variables. - fixed: 'none' as class type must map to the real null pointer so that it won't get rejected by the stricter type checks. - added handling for member function calls to zcc_compile.cpp. - fixed: FxMemberFunctionCall may not delete the self expression if it gets passed on to the actual function call.
2016-10-22 17:43:53 +00:00
{
auto ema = static_cast<ZCC_ExprMemberAccess *>(fcall->Function);
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
return new FxMemberFunctionCall(ConvertNode(ema->Left), ema->Right, ConvertNodeList(args, fcall->Parameters), *ast);
- fixed a conversion warning with the pointer-type AActor fields. - added master and tracer to the list of exported variables. - fixed: 'none' as class type must map to the real null pointer so that it won't get rejected by the stricter type checks. - added handling for member function calls to zcc_compile.cpp. - fixed: FxMemberFunctionCall may not delete the self expression if it gets passed on to the actual function call.
2016-10-22 17:43:53 +00:00
}
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
case AST_ExprBinary:
// Array syntax for randoms. They are internally stored as ExprBinary with both an identifier on the left and right side.
if (fcall->Function->Operation == PEX_ArrayAccess)
{
auto binary = static_cast<ZCC_ExprBinary *>(fcall->Function);
if (binary->Left->NodeType == AST_ExprID && binary->Right->NodeType == AST_ExprID)
{
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
return new FxFunctionCall(static_cast<ZCC_ExprID *>(binary->Left)->Identifier, static_cast<ZCC_ExprID *>(binary->Right)->Identifier, ConvertNodeList(args, fcall->Parameters), *ast);
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
}
}
// fall through if this isn't an array access node.
default:
Error(fcall, "Invalid function identifier");
return new FxNop(*ast); // return something so that the compiler can continue.
}
break;
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
}
added class pointer casts. Due to grammar problems the type has to be put into parentheses to get the class token out of the global parsing namespace: class<Actor> myclass = (class<Actor>)(GetClass());
2016-11-17 19:31:53 +00:00
case AST_ClassCast:
{
auto cc = static_cast<ZCC_ClassCast *>(ast);
if (cc->Parameters == nullptr || cc->Parameters->SiblingNext != cc->Parameters)
{
Error(cc, "Class type cast requires exactly one parameter");
return new FxNop(*ast); // return something so that the compiler can continue.
}
auto cls = PClass::FindClass(cc->ClassName);
if (cls == nullptr)
{
Error(cc, "Unknown class %s", FName(cc->ClassName).GetChars());
return new FxNop(*ast); // return something so that the compiler can continue.
}
return new FxClassPtrCast(cls, ConvertNode(cc->Parameters));
}
- added static constant arrays. At the moment they can only be defined inside functions due to lack of dedicated storage inside classes for static data. - added new VM instructions to access the constant tables with a variable index. - refactored VMFunctionBuilder's constant tables so that they are not limited to one entry per value. While this works fine for single values, it makes it impossible to store constant arrays in here.
2016-11-20 17:00:37 +00:00
case AST_StaticArrayStatement:
{
auto sas = static_cast<ZCC_StaticArrayStatement *>(ast);
PType *ztype = DetermineType(ConvertClass, sas, sas->Id, sas->Type, false, false);
FArgumentList args;
ConvertNodeList(args, sas->Values);
// This has to let the code generator resolve the constants, not the Simplifier, which lacks all the necessary type info.
return new FxStaticArray(ztype, sas->Id, args, *ast);
}
added class pointer casts. Due to grammar problems the type has to be put into parentheses to get the class token out of the global parsing namespace: class<Actor> myclass = (class<Actor>)(GetClass());
2016-11-17 19:31:53 +00:00
- implemented class member access for variables. (Struct members and functions not done yet.)
2016-10-21 15:41:39 +00:00
case AST_ExprMemberAccess:
{
auto memaccess = static_cast<ZCC_ExprMemberAccess *>(ast);
return new FxMemberIdentifier(ConvertNode(memaccess->Left), memaccess->Right, *ast);
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
case AST_FuncParm:
{
auto fparm = static_cast<ZCC_FuncParm *>(ast);
- implemented named arguments. - fixed flag CVAR access. As it turned out, OP_LBIT is a bit messy to set up properly when accessing integers that may or may not be big endian, so it now uses a shift and bit masking to do its work. - used the SpawnPlayerMissile call in A_FireBFG to test named arguments.
2016-11-21 00:32:01 +00:00
auto node = ConvertNode(fparm->Value);
if (fparm->Label != NAME_None) node = new FxNamedNode(fparm->Label, node, *ast);
return node;
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
}
- added handling for all the builtins supported by DECORATE to FxFunctionCall. For the random functions this class only handles the default-RNG version. The one with an explicit RNG needs to be done separately because the parser produces different output for them.
2016-10-15 22:12:33 +00:00
case AST_ExprID:
{
auto id = static_cast<ZCC_ExprID *>(ast);
return new FxIdentifier(id->Identifier, *ast);
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
case AST_ExprConstant:
{
auto cnst = static_cast<ZCC_ExprConstant *>(ast);
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
if (cnst->Type->IsA(RUNTIME_CLASS(PName)))
{
return new FxConstant(FName(ENamedName(cnst->IntVal)), *ast);
}
else if (cnst->Type->IsA(RUNTIME_CLASS(PInt)))
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
return new FxConstant(cnst->IntVal, *ast);
}
- converted the rest of Heretic's actors. - fixed: Boolean constants were not properly handled.
2016-10-17 10:58:23 +00:00
else if (cnst->Type->IsA(RUNTIME_CLASS(PBool)))
{
return new FxConstant(!!cnst->IntVal, *ast);
}
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
else if (cnst->Type->IsA(RUNTIME_CLASS(PFloat)))
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
return new FxConstant(cnst->DoubleVal, *ast);
}
- added FindClassMemberFunction which retrieves a function symbol and performs some verification. - removed Self parameter from FxFunctionCall. Actual member function calls through an object require quite different handling so lumping these two together makes no sense. - added a workaround to deal with ACS_NamedExecuteWithResult to both the compiler and FindClassMemberFunction. The way the ZScript compiler sets this up means that it will call the builtin, not the actual action function, so the parser needs to do some explicit check to get past the same-named action function. - pass a proper self pointer to FxActionSpecial. Although it's still not being used, propagating design shortcuts through several function levels is a very, very bad idea.
2016-10-15 18:16:44 +00:00
else if (cnst->Type->IsA(RUNTIME_CLASS(PString)))
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
return new FxConstant(*cnst->StringVal, *ast);
}
- added null pointer. - fixed: FxMemberIdentifier checked for ClassPointers instead of object pointers to resolve the left hand side of the expression. - allow comparison of pointers.
2016-10-21 22:50:04 +00:00
else if (cnst->Type == TypeNullPtr)
{
return new FxConstant(*ast);
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
else
{
// can there be other types?
- converted the rest of Heretic's actors. - fixed: Boolean constants were not properly handled.
2016-10-17 10:58:23 +00:00
Error(cnst, "Unknown constant type %s", cnst->Type->DescriptiveName());
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
return new FxConstant(0, *ast);
}
}
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case AST_ExprUnary:
{
auto unary = static_cast<ZCC_ExprUnary *>(ast);
auto operand = ConvertNode(unary->Operand);
auto op = unary->Operation;
switch (op)
{
case PEX_PostDec:
case PEX_PostInc:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxPostIncrDecr(operand, Pex2Tok[op]);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_PreDec:
case PEX_PreInc:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxPreIncrDecr(operand, Pex2Tok[op]);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_Negate:
return new FxMinusSign(operand);
case PEX_AntiNegate:
return new FxPlusSign(operand);
case PEX_BitNot:
return new FxUnaryNotBitwise(operand);
case PEX_BoolNot:
return new FxUnaryNotBoolean(operand);
case PEX_SizeOf:
case PEX_AlignOf:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxSizeAlign(operand, Pex2Tok[op]);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
default:
assert(0 && "Unknown unary operator."); // should never happen
Error(unary, "Unknown unary operator ID #%d", op);
return new FxNop(*ast);
}
break;
}
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
case AST_ExprBinary:
{
auto binary = static_cast<ZCC_ExprBinary *>(ast);
auto left = ConvertNode(binary->Left);
auto right = ConvertNode(binary->Right);
auto op = binary->Operation;
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
auto tok = Pex2Tok[op];
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
switch (op)
{
case PEX_Add:
case PEX_Sub:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxAddSub(tok, left, right);
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
case PEX_Mul:
case PEX_Div:
case PEX_Mod:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxMulDiv(tok, left, right);
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
- implemented '**' (power) operator. To ensure reliability, acustom 'pow' function will be used to calculate it. - fixed: FxBinary::ResolveLR' check for numeric operations was incomplete. Like far too many other places it just assumed that everything with ValueType->GetRegType() == REGT_INT is a numeric type, but for names this is not the case.
2016-10-17 13:17:48 +00:00
case PEX_Pow:
return new FxPow(left, right);
- added handling for shift and bit operators. - added the '>>>' (unsigned shift) operator. Although, with unsigned numbers available, this is technically not really needed, DECORATE supports this so ZScript should, too, if only for the benefit of making conversion tools easier to handle.
2016-10-17 13:52:29 +00:00
case PEX_LeftShift:
case PEX_RightShift:
case PEX_URightShift:
- cleanup of the remaining FxBinary operators. - changed FxCompareEq with strings and other types that can be cast to from a string always convert the string to the other type before comparing.
2016-11-18 13:19:55 +00:00
return new FxShift(tok, left, right);
- added handling for shift and bit operators. - added the '>>>' (unsigned shift) operator. Although, with unsigned numbers available, this is technically not really needed, DECORATE supports this so ZScript should, too, if only for the benefit of making conversion tools easier to handle.
2016-10-17 13:52:29 +00:00
case PEX_BitAnd:
case PEX_BitOr:
case PEX_BitXor:
- cleanup of the remaining FxBinary operators. - changed FxCompareEq with strings and other types that can be cast to from a string always convert the string to the other type before comparing.
2016-11-18 13:19:55 +00:00
return new FxBitOp(tok, left, right);
- added handling for shift and bit operators. - added the '>>>' (unsigned shift) operator. Although, with unsigned numbers available, this is technically not really needed, DECORATE supports this so ZScript should, too, if only for the benefit of making conversion tools easier to handle.
2016-10-17 13:52:29 +00:00
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_BoolOr:
case PEX_BoolAnd:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxBinaryLogical(tok, left, right);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_LT:
case PEX_LTEQ:
case PEX_GT:
case PEX_GTEQ:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxCompareRel(tok, left, right);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_EQEQ:
case PEX_NEQ:
- implemented ~== operator. Turned out this was really simple because the functionality was already there.
2016-10-22 07:31:37 +00:00
case PEX_APREQ:
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
return new FxCompareEq(tok, left, right);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
- changed assignment operators to be expressions, like they are in C and DECORATE. This also means that for now Lua-style multi-assignments are disabled, those should be easy to enable by making some changes to the assignment_statement grammar so that it doesn't recognize single assignments, but for now this is low priority because it adds a significant amount of complexity to do this right with functions that have multiple return values.
2016-10-21 10:22:42 +00:00
case PEX_Assign:
return new FxAssign(left, right);
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
case PEX_AddAssign:
case PEX_SubAssign:
return ModifyAssign(new FxAddSub(tok, new FxAssignSelf(*ast), right), left);
case PEX_MulAssign:
case PEX_DivAssign:
case PEX_ModAssign:
return ModifyAssign(new FxMulDiv(tok, new FxAssignSelf(*ast), right), left);
case PEX_LshAssign:
case PEX_RshAssign:
case PEX_URshAssign:
- cleanup of the remaining FxBinary operators. - changed FxCompareEq with strings and other types that can be cast to from a string always convert the string to the other type before comparing.
2016-11-18 13:19:55 +00:00
return ModifyAssign(new FxShift(tok, new FxAssignSelf(*ast), right), left);
- implemented all assign/modify operators. - use a table to translate from PEX to tokens to make the code easier to read.
2016-10-21 12:18:31 +00:00
case PEX_AndAssign:
case PEX_OrAssign:
case PEX_XorAssign:
- cleanup of the remaining FxBinary operators. - changed FxCompareEq with strings and other types that can be cast to from a string always convert the string to the other type before comparing.
2016-11-18 13:19:55 +00:00
return ModifyAssign(new FxBitOp(tok, new FxAssignSelf(*ast), right), left);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_LTGTEQ:
- implemented the '<>=' operator - more as a VM coding exercise than for its usefulness, though.
2016-10-21 14:35:07 +00:00
return new FxLtGtEq(left, right);
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
case PEX_ArrayAccess:
return new FxArrayElement(left, right);
- handle all unary operators. - handle all binary operators which are already implemented in the code generator. - implemented sizeof/alignof operators in code generator. - rewrote RequestAddress so that its return value is not the writability of an address but the mere existence. Also changed it to not output errors itself because those cannot be dealt with by the calling function.
2016-10-17 18:33:35 +00:00
case PEX_CrossProduct:
case PEX_DotProduct:
- implemented dot and cross products (not that I'd ever expect these to become useful in a game like Doom...)
2016-10-29 08:16:00 +00:00
return new FxDotCross(tok, left, right);
- changed assignment operators to be expressions, like they are in C and DECORATE. This also means that for now Lua-style multi-assignments are disabled, those should be easy to enable by making some changes to the assignment_statement grammar so that it doesn't recognize single assignments, but for now this is low priority because it adds a significant amount of complexity to do this right with functions that have multiple return values.
2016-10-21 10:22:42 +00:00
- implemented the 'is' operator (i.e. class type check.)
2016-10-29 23:05:56 +00:00
case PEX_Is:
return new FxTypeCheck(left, right);
- implemented dot and cross products (not that I'd ever expect these to become useful in a game like Doom...)
2016-10-29 08:16:00 +00:00
case PEX_Concat:
- implemented the string concatenation operator '..'. This is capable of stringifying all of the common types for output. - gave OP_CONCAT some sane semantics. The way this was defined, by specifying the source operands as a range of registers instead of a pair like everything else made it completely useless for the task at hand. - changed formatting for floats to %.5f which for normal output in a game makes more sense. For special cases there should be a special formatting function for ints and floats that can do more specialized conversions.
2016-11-18 16:44:25 +00:00
return new FxConcat(left, right);
- changed assignment operators to be expressions, like they are in C and DECORATE. This also means that for now Lua-style multi-assignments are disabled, those should be easy to enable by making some changes to the assignment_statement grammar so that it doesn't recognize single assignments, but for now this is low priority because it adds a significant amount of complexity to do this right with functions that have multiple return values.
2016-10-21 10:22:42 +00:00
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
default:
I_Error("Binary operator %d not implemented yet", op);
}
break;
}
- handle trinary conditional statement.
2016-10-17 19:43:14 +00:00
case AST_ExprTrinary:
{
auto trinary = static_cast<ZCC_ExprTrinary *>(ast);
auto condition = ConvertNode(trinary->Test);
auto left = ConvertNode(trinary->Left);
auto right = ConvertNode(trinary->Right);
return new FxConditional(condition, left, right);
}
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
Renamed "VectorInitializer" to "VectorValue" This was really confusing for me as this is an actual vector "value" rather than an "initializer"
2016-10-30 03:35:01 +00:00
case AST_VectorValue:
- implemented local vector variables. Currently only the definition plus initial assignment works. - removed all vector4 handling that had already been added, now that this type can no longer be defined.
2016-10-28 13:15:30 +00:00
{
Renamed "VectorInitializer" to "VectorValue" This was really confusing for me as this is an actual vector "value" rather than an "initializer"
2016-10-30 03:35:01 +00:00
auto vecini = static_cast<ZCC_VectorValue *>(ast);
- implemented local vector variables. Currently only the definition plus initial assignment works. - removed all vector4 handling that had already been added, now that this type can no longer be defined.
2016-10-28 13:15:30 +00:00
auto xx = ConvertNode(vecini->X);
auto yy = ConvertNode(vecini->Y);
auto zz = ConvertNode(vecini->Z);
Renamed "VectorInitializer" to "VectorValue" This was really confusing for me as this is an actual vector "value" rather than an "initializer"
2016-10-30 03:35:01 +00:00
return new FxVectorValue(xx, yy, zz, *ast);
- implemented local vector variables. Currently only the definition plus initial assignment works. - removed all vector4 handling that had already been added, now that this type can no longer be defined.
2016-10-28 13:15:30 +00:00
}
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
case AST_LocalVarStmt:
{
auto loc = static_cast<ZCC_LocalVarStmt *>(ast);
auto node = loc->Vars;
FxSequence *list = new FxSequence(*ast);
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
PType *ztype = DetermineType(ConvertClass, node, node->Name, loc->Type, true, false);
if (loc->Type->ArraySize != nullptr)
{
ztype = ResolveArraySize(ztype, loc->Type->ArraySize, &ConvertClass->Symbols);
}
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
do
{
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
PType *type;
if (node->ArraySize != nullptr)
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
{
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
type = ResolveArraySize(ztype, node->ArraySize, &ConvertClass->Symbols);
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
}
else
{
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
type = ztype;
}
FxExpression *val;
if (node->InitIsArray)
{
Error(node, "Compound initializer not implemented yet");
val = nullptr;
}
else
{
val = node->Init ? ConvertNode(node->Init) : nullptr;
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
}
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
list->Add(new FxLocalVariableDeclaration(type, node->Name, val, 0, *node)); // todo: Handle flags in the grammar.
- split up FxCompoundStatement into the old FxSequence and a new subclass. FxSequence is just a sequence of expressions, while FxCompoundStatement is the one that actually implements handling of local variables. The split was done so that ZCCCompiler::ConvertNode can return multiple statements as one object and for that FxCompoundStatement is not suitable. - added conversion of local variable declarations. This is still untested
2016-10-19 23:09:35 +00:00
node = static_cast<decltype(node)>(node->SiblingNext);
} while (node != loc->Vars);
return list;
}
- implemented super calls and proper dispatch of scripted virtual overrides for Destroy.
2016-11-11 20:52:08 +00:00
case AST_Expression:
{
auto ret = static_cast<ZCC_Expression *>(ast);
if (ret->Operation == PEX_Super)
{
return new FxSuper(*ast);
}
break;
}
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
case AST_ExpressionStmt:
return ConvertNode(static_cast<ZCC_ExpressionStmt *>(ast)->Expression);
case AST_ReturnStmt:
{
auto ret = static_cast<ZCC_ReturnStmt *>(ast);
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
FArgumentList args;
ConvertNodeList(args, ret->Values);
if (args.Size() == 0)
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
{
return new FxReturnStatement(nullptr, *ast);
}
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
else if (args.Size() == 1)
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
{
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
auto arg = args[0];
args[0] = nullptr;
return new FxReturnStatement(arg, *ast);
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
}
else
{
Error(ast, "Return with multiple values not implemented yet.");
return new FxReturnStatement(nullptr, *ast);
}
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
}
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
- process most of the remaining statement types. - extend assignment operations to handle local variables (untested.) - make the implicit function arguments read only.
2016-10-20 23:12:54 +00:00
case AST_BreakStmt:
case AST_ContinueStmt:
return new FxJumpStatement(ast->NodeType == AST_BreakStmt ? TK_Break : TK_Continue, *ast);
case AST_IfStmt:
{
auto iff = static_cast<ZCC_IfStmt *>(ast);
return new FxIfStatement(ConvertNode(iff->Condition), ConvertNode(iff->TruePath), ConvertNode(iff->FalsePath), *ast);
}
case AST_IterationStmt:
{
auto iter = static_cast<ZCC_IterationStmt *>(ast);
if (iter->CheckAt == ZCC_IterationStmt::End)
{
assert(iter->LoopBumper == nullptr);
return new FxDoWhileLoop(ConvertNode(iter->LoopCondition), ConvertNode(iter->LoopStatement), *ast);
}
else if (iter->LoopBumper != nullptr)
{
return new FxForLoop(nullptr, ConvertNode(iter->LoopCondition), ConvertNode(iter->LoopBumper), ConvertNode(iter->LoopStatement), *ast);
}
else
{
return new FxWhileLoop(ConvertNode(iter->LoopCondition), ConvertNode(iter->LoopStatement), *ast);
}
}
- implemented class member access for variables. (Struct members and functions not done yet.)
2016-10-21 15:41:39 +00:00
// not yet done
case AST_SwitchStmt:
- added switch/case processing. Right now it is just a sequence of test/jmp instructions. It may make sense to add a special opcode that can perform the comparisons natively but that's an option for later. - added a TESTN instruction. This is like TEST but negates the operand. This was added to avoid flooding the constant table with too many case labels. With TEST and TESTN combined, all numbers between -65535 and 65535 can be kept entirely inside the instruction. Numbers outside this range still use a BEQ instruction.
2016-10-23 10:00:25 +00:00
{
auto swtch = static_cast<ZCC_SwitchStmt *>(ast);
if (swtch->Content->NodeType != AST_CompoundStmt)
{
Error(ast, "Expecting { after 'switch'");
return new FxNop(*ast); // allow compiler to continue looking for errors.
}
else
{
// The switch content is wrapped into a compound statement which needs to be unraveled here.
auto cmpnd = static_cast<ZCC_CompoundStmt *>(swtch->Content);
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
FArgumentList args;
return new FxSwitchStatement(ConvertNode(swtch->Condition), ConvertNodeList(args, cmpnd->Content), *ast);
- added switch/case processing. Right now it is just a sequence of test/jmp instructions. It may make sense to add a special opcode that can perform the comparisons natively but that's an option for later. - added a TESTN instruction. This is like TEST but negates the operand. This was added to avoid flooding the constant table with too many case labels. With TEST and TESTN combined, all numbers between -65535 and 65535 can be kept entirely inside the instruction. Numbers outside this range still use a BEQ instruction.
2016-10-23 10:00:25 +00:00
}
}
- implemented class member access for variables. (Struct members and functions not done yet.)
2016-10-21 15:41:39 +00:00
- added switch/case processing. Right now it is just a sequence of test/jmp instructions. It may make sense to add a special opcode that can perform the comparisons natively but that's an option for later. - added a TESTN instruction. This is like TEST but negates the operand. This was added to avoid flooding the constant table with too many case labels. With TEST and TESTN combined, all numbers between -65535 and 65535 can be kept entirely inside the instruction. Numbers outside this range still use a BEQ instruction.
2016-10-23 10:00:25 +00:00
case AST_CaseStmt:
{
auto cases = static_cast<ZCC_CaseStmt *>(ast);
- scriptified A_Saw. - implemented multiple-return-value assignment. Due to some grammar conflicts the originally intended Lua-inspired syntax of 'a, b = Function()' could not be done, so it's '[a, b] = Function()'
2016-11-19 23:25:38 +00:00
return new FxCaseStatement(ConvertNode(cases->Condition), *ast);
- added switch/case processing. Right now it is just a sequence of test/jmp instructions. It may make sense to add a special opcode that can perform the comparisons natively but that's an option for later. - added a TESTN instruction. This is like TEST but negates the operand. This was added to avoid flooding the constant table with too many case labels. With TEST and TESTN combined, all numbers between -65535 and 65535 can be kept entirely inside the instruction. Numbers outside this range still use a BEQ instruction.
2016-10-23 10:00:25 +00:00
}
- process most of the remaining statement types. - extend assignment operations to handle local variables (untested.) - make the implicit function arguments read only.
2016-10-20 23:12:54 +00:00
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
case AST_CompoundStmt:
{
auto x = new FxCompoundStatement(*ast);
auto compound = static_cast<ZCC_CompoundStmt *>(ast);
auto node = compound->Content;
if (node != nullptr) do
{
x->Add(ConvertNode(node));
node = static_cast<decltype(node)>(node->SiblingNext);
} while (node != compound->Content);
return x;
}
- scriptified A_Saw. - implemented multiple-return-value assignment. Due to some grammar conflicts the originally intended Lua-inspired syntax of 'a, b = Function()' could not be done, so it's '[a, b] = Function()'
2016-11-19 23:25:38 +00:00
case AST_AssignStmt:
{
auto ass = static_cast<ZCC_AssignStmt *>(ast);
FArgumentList args;
ConvertNodeList(args, ass->Dests);
assert(ass->Sources->SiblingNext == ass->Sources); // right side should be a single function call - nothing else
if (ass->Sources->NodeType != AST_ExprFuncCall)
{
// don't let this through to the code generator. This node is only used to assign multiple returns of a function to more than one variable.
Error(ass, "Right side of multi-assignment must be a function call");
return new FxNop(*ast); // allow compiler to continue looking for errors.
}
return new FxMultiAssign(args, ConvertNode(ass->Sources), *ast);
}
- Fixed more GCC/Clang warnings.
2016-11-26 11:45:17 +00:00
default:
break;
- added processing of compound statements to the compiler. This means that anonymous functions without control statements are generating code now. - added local variable declarations to the code generator. This is not tested yet, that will come with the next commit.
2016-10-19 17:05:48 +00:00
}
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
// only for development. I_Error is more convenient here than a normal error.
I_Error("ConvertNode encountered unsupported node of type %d", ast->NodeType);
return nullptr;
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
}
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
FArgumentList &ZCCCompiler::ConvertNodeList(FArgumentList &args, ZCC_TreeNode *head)
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
if (head != nullptr)
- converted a few more DECORATE files. - started with the AST converter. So far it only deals with direct function calls with simple constants as parameters. - added an error condition for the defaults block to get rid of some asserts.
2016-10-13 22:40:20 +00:00
{
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
auto node = head;
do
{
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
args.Push(ConvertNode(node));
- implemented array-style syntax for Random calls. - implemented handling of the basic math operators so that heretic/beast.txt can be processed. This is working, aside from still needing the type casts to properly transform the strings to class pointers.
2016-10-16 08:59:12 +00:00
node = node->SiblingNext;
} while (node != head);
}
use a memory arena for allocating code generation nodes. - Since the number of small allocations here is extremely high this will help a lot to prevent fragmentation and since most nodes are collected up front and this is done when no large resources are being loaded it won't cause heap spikes. let Emit methods delete FxExpression arrays when they are done. - For some reason the deletion process does not work 100%, there are always some nodes left behind and so far I haven't found them. This ensures that these arrays do not live any longer than needed.
2016-11-10 14:13:31 +00:00
return args;
- Fixed more GCC/Clang warnings.
2016-11-26 11:45:17 +00:00
}
Reference in a new issue Copy permalink