Stopping a source in the middle of playback can causing undesirable "popping"
from a sudden change in amplitude. Setting the source gain to 0 gives OpenAL
the opportunity to ramp the source's output to 0, where it can be safely
stopped without causing a sudden change in amplitude.
This isn't supported with the OpenAL backend, and instead of trying to make a
workaround for it, a better approach would probably be to make a FileReader
implementation that handles URLs.
Added fallback to generic Cocoa implementation if it's not available (i.e. without OpenGL Core Profile support)
Value of vid_glswfb CVAR is now saved to configuration file
This reverts commit 545241aa06.
These commits are WIP stuff that should not be in the master branch. A work branch has been created to finalize this code.
src/polyrenderer/drawers/poly_triangle.cpp:212:20: error: ‘CPU’ was not declared in this scope
src/polyrenderer/drawers/screen_triangle.cpp:955:6: error: ‘CPU’ was not declared in this scope
src/sound/oalsound.cpp:1288: error: non-constant-expression cannot be narrowed from type 'uint32_t' (aka 'unsigned int') to 'ALint' (aka 'int') in initializer list [-Wc++11-narrowing]
Allocation of 810 FState objects (32400 bytes in total) could be performed in MSVC x64 build but worked in 32-bit version
This fixes https://mantis.zdoom.org/view.php?id=602
- added arrays to the config to hold entries for the softsynths' config files. This is not active yet, but will later be used to give the user a list of config options instead of having to type it by hand.
The needed headers are now included in the repo, which for these libraries is possible thanks to a stable ABI (at least on Windows, the other platforms still need to be checked but the headers only add, never remove or change existing content.)
The big advantage of this setup is that it allows building the project on Windows without any necessary setup - all that needs to be provided is the DLLs from the binary package.
This still requires some fixes for macOS and Linux. On MacOS the proper library names are missing and the ones for Linux are not verified. Both platforms should work, though, if the dynamic loading is disabled.
For some files that had the Doom Source license attached but saw heavy external contributions over the years I added a special note to license all original ZDoom code under BSD.
src/sound/oplsynth/musicblock.cpp:3:10: fatal error: 'muslib.h' file not found
src/sound/oplsynth/oplio.cpp:410:12: error: use of undeclared identifier 'cos'
src/sound/oplsynth/oplio.cpp:410:41: error: use of undeclared identifier 'sin'
This was very poorly done without ever addressing the issues a composite render style can bring, it merely dealt with the known legacy render styles.
The same, identical code was also present in two different places.
The oversight that AlterWeaponSprite overrode even forced styles was also fixed.
OpenGL is not implemented yet but with the problems eliminated should be doable now.
This is to ensure that the Class pointer can be set right on creation. ZDoom had always depended on handling this lazily which poses some problems for the VM.
So now there is a variadic Create<classtype> function taking care of that, but to ensure that it gets used, direct access to the new operator has been blocked.
This also neccessitated making DArgs a regular object because they get created before the type system is up. Since the few uses of DArgs are easily controllable this wasn't a big issue.
- did a bit of optimization on the bots' decision making whether to pick up a health item or not.
The global variable holding a pointer to this thinker should be a weak reference to the instance in the thinker chain, there is no need to mark this global variable, as the thinker's lifetime is only determined by the thinker chain.
- committed a few Posix related file the last commit missed.
I have no idea why they were even in there, as they intentionally circumvented all GC related features - they declared themselves fixed if prone to getting collected, they all used OF_YesReallyDelete when destroying themselves and they never used any of the object creation or RTTI features, aside from a single assert in V_Init2.
Essentially they were a drag on the system and OF_YesReallyDelete was effectively added just to deal with the canvases which were DObjects but not supposed to behave like them in the first place.
Let's use inline checkers in PType instead of constantly having to do clumsy IsKindOf checks etc. Once complete this also means that the types can be taken out of the class hierarchy, freeing up some common names.
src/p_pspr.cpp:363:37: warning: more '%' conversions than data arguments [-Wformat]
src/gl/textures/gl_texture.cpp:845:21: warning: comparison of integers of different signs: 'int' and 'unsigned int' [-Wsign-compare]
Fixed bunch of compilation errors:
cannot pass non-trivial object of type 'FString' to variadic method; expected type from format string was 'char *' [-Wnon-pod-varargs]
Fixed linker erorr:
g_doomedmap.cpp.o: In function `InitActorNumsFromMapinfo()':
src/g_doomedmap.cpp: undefined reference to `PClass::FindActor(FName)'
Combining these two groups of data has been the cause of many hard to detect errors because it allowed liberal casting between types that are used for completely different things.
Having these two types related can cause problems with detection in the compiler because for some parts they need quite different handling.
Common handling for the fields has been moved into PSymbolTable but overall redundancy was quite minor as both types share surprisingly little functionality.
A pointer to an object has quite different semantics so let's do this with proper virtual inheritance. This should allow to simplify a lot of pointer related checks in the compiler.
- fixed: IsVisibleToPlayer and optimized it a bit more.
- moved SourceLumpName to PClass, so that it can also be used for non-actors (there's a lot of non-Actor classes already.)
- separated the serializer for PClassPointer from PPointer. Even though not doable, a pointer to a class type is something entirely different than a class pointer with a restriction so each should handle its own case.
- added a few access functions for FActorInfo variables.
With PClassActor now empty the class descriptors can finally be converted back to static data outside the class hierarchy, like they were before the scripting merge, and untangle the game data from VM internals.
For these fields maps have no advantage. Linearly searching a small array with up to 10 entries is nearly always faster than generating a hash for finding the entry in the map.
- some optimization of access to OwnedStates in old DECORATE.
- consolidate all places that print a state name into a subfunction.
- allocate states from the ClassDataAllocator memory arena. States do not need to be freed separately from the rest of the static class data.
This reinstates the old FActorInfo as part of the meta data a class can have so that the class descriptor itself can be freed from any data not directly relevant for managing the class's type information.
This is not needed anymore because classes do not need to be replaced. The only reason this was implemented was the original design with the class descriptors taking on all the metadata themselves.
This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere.
Note that this commit does not remove the tags, it only discontinues their use.
See https://forum.drdteam.org/viewtopic.php?t=7588
Processing order is now the same as in Chocolate Doom
prBoom+ loads separate files after all WAD lumps though
This makes sense but would change loading sequence existed in ZDoom for years
This is somewhat brute-force thanks to the surprising lack of good documentation for the Ogg headers. The only other option would have been some rather bloated library for a function that should be 25-30 lines at most.
The idea is to have more control on the game side instead of dealing with these formats in the backend, which was done for FMod because it already had the decoders implemented.
However, with OpenAL this setup makes no sense and only complicates future extensions that can be better handled at a higher level.
No more error when running with +map command line parameter with classic HUD:
> VM execution aborted: Attempt to draw to screen outside a draw function
> Called from BaseStatusBar.DrawImage [Native]
> Called from DoomStatusBar.DrawFullScreenStuff at gzdoom.pk3:zscript/statusbar/doom_sbar.txt, line 140
> Called from DoomStatusBar.Draw at gzdoom.pk3:zscript/statusbar/doom_sbar.txt, line 41
- all 5 settings affected by uiscale have been changed to have the exact same semantics: -1, if supported means special scaling, this is available for HUD and status bar, 0 means to use uiscale, any larger value is a direct scaling factor.
- scaling is cut off when the factor is larger than screenwidth/320 or screenheight/200 because anything larger will definitely not fit.
- a lot of code has been cleaned up and consolidated. Especially the message code had an incredible amount of redundancy.
- all scaling options have been moved into a submenu. This menu is not complete, though - it still requires a special menu widget to convey the intended information without confusing the user.
- activated the RenderOverlay event, now that it can be called from the correct spot, i.e. right after the top level HUD messages are drawn. The system's status output will still be drawn on top of them.
- changed the effect spawn prevention of the Hexen flame strike weapon and reverted the attempt to fix this in FastProjectile.
This cannot be fixed in the base class, which was doing everything right. It's the flame missile that was doing undefined things by stopping its movement without clearing its missile flag. This cannot work because missiles are given some minimal forced velocity to ensure collision detection and any attempt to address this without clearing the missile flag is doomed to fail.
This allows using the UI scale or its own value, like all other scaling values.
In addition there is a choice between preserving equal pixel size or aspect ratio because the squashed non-corrected versions tend to look odd, but since proper scaling requires ununiform pixel sizes it is an option.
- changed how status bar sizes are being handled.
This has to recalculate all scaling and positioning factors, which can cause problems if the drawer leaves with some temporary values that do not reflect the status bar as a whole.
Changed it so that the status bar stores the base values and restores them after drawing is complete.
This would cut off overlong names and the handling for status bars with protruding elements was far too simplistic and worse, making assumptions based on game mode.
It now uses a virtual function to query the status bar itself for returning this information so it can be overridden and uses V_BreakLines to split the text if it is wider than the display.
This is necessary because the hardware accelerated renderers will hide the problem, but with pure software rendering to a locked hardware surface, like DirectDraw can result in a crash.
Note that ANY mod that gets caught in this did something wrong!
Currently this is only being used for draw operations that are not automap related, i.e. DrawLine, DrawPixel and FillSimplePoly are not subjected to it.
All our continuous integration targets have no problems with C99 isnan() macro but on Ubuntu 16.04 compilation fails
It appeared that some implementation of C++ Standard Library may undefine bunch of C macros to avoid conflicts with own declarations
- finished work on the Doom status bar. I also took the opportunity to fix the layout of the inventory bar which is a bit broken in SBARINFO.
- tuned the selection rules for deciding what creates the status bar, so that the most recent definition that can be found is chosen.
- got rid of the image list in the Doom status bar. The cost of the texture lookup is mostly irrelevant here so clearer and shorter code is preferrable.
- moved the box fitting code from DrawTexture into the native function to have all coordinate calculations in one place which is necessary to implement proper alignment default handling. Without higher level functions altering positioning the default can be set to automatic alignment determination, i.e. the value's sign decides where something is placed. Of course for special cases this can be overridden.
- use ANIMDEFS to animate the inventory arrow,
Fullscreen HUD done with the exception of key and inventory bar. I also used the opportunity to make it a bit more resistant against badly designed inventory icons.
- better handling of ForceScale for the fullscreen HUD that doesn't mess around with CVARs.
- moved the mug shot into the status bar, because this is global state that needs to be shared between different pieces of code which want to display a mug shot.
- SBARINFO should work off the current status bar settings instead of the ones stored in its script object
Note that there is no direct access, all this exposes is a single function to get the current face's texture which then can be drawn using the existing functions.
Note that the pop screens are special because they are not subject to scaling - they will always be drawn with the current resolutions clean scale. As a result they cannot use the HUD drawers but instead continue to use the low level draw functions directly.
- replaced Inventory.DrawPowerup with a GetPowerupIcon method so that the calling code can handle the drawing and apply its own rules. This was a major design flaw of allowing the inventory items to handle the drawing themselves, because they were unable to adjust to different HUD frontends. Note that any mod that overrides DrawPowerup will not draw any icon that expects to be handled that way!
- the alternative HUD now has its own, separate drawer that obeys the AltHUD's rules, and not the ones of the normal fullscreen HUD.
- the standard drawer has been scriptified as a virtual function.
- Both drawers now handle positioning of the icon inside its assigned box themselves instead of trusting the powerup item to do it correctly.
- DTA_HUDRules and Screen.DrawHUDTexture are to be considered deprecated because both do not integrate into the redesigned HUD code.
- made bluramount also into a gameinfo option
- negative gl_menu_blur cvar now uses gameinfo option, 0 disables it
- removed gl_menu_blur_enabled since gl_menu_blur==0 does that anyway
- made gl_menu_blur default to -1 to use gameinfo option
- add default gameinfo bluramount options
- merged BaseStatusBar and CustomStatusBar back together.
Since the low level draw functions are better done in native code for both performance and debuggability the split has become pointless.
- decided to ditch the widget system I had started to lay out. As it turns out that would make things far more complicated and slower than they need to be.
This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string.
This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit.
This is easily offset by the code savings and reduced maintenance needs elsewhere.
Note that the Strife status bar does not draw the health bars yet. I tried to replace the hacky custom texture with a single fill operation but had to find out that all the coordinate mangling for the status bar is being done deep in the video code. This needs to be fixed before this can be made to work.
Currently this is not usable in mods because they cannot initialize custom status bars yet.
Using I_MSTime is not precise enough, because some camera textures can be done quicker. It was pointless anyway trying to make this multithreading-safe, the entire caching idea here makes no sense if two clippers can simultaneously work on the same level data without changing the memory organization and rendering it ineffective.
It used the expression's value type, but needs to use the variable's, which can be different when the assignment is synthesized from a builtin function.
- keep string registers which are being used as function parameters allocated until after the function call returns. This is for allowing to pass strings by reference which would avoid some costly constructor/destructor loops in the call instruction.
Back to Saturn X E1: Get Out Of My Stations (https://www.doomworld.com/idgames/levels/doom2/megawads/btsx_e1) uses tabs instead of spaces and bunch of 'Need data after par' warnings were printed during loading
PrBoom supports arbitrary space characters too
src/gl/scene/gl_clipper.h:150:23: warning: comparison of integers of different signs: 'int' and 'unsigned int' [-Wsign-compare]
src/gl/dynlights/gl_aabbtree.cpp:137:24: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:137:34: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:137:44: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:139:6: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:139:30: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:139:54: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:142:6: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:143:3: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:144:3: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_aabbtree.cpp:167:6: warning: using integer absolute value function 'abs' when argument is of floating point type [-Wabsolute-value]
src/gl/dynlights/gl_shadowmap.cpp:163:31: warning: '&&' within '||' [-Wlogical-op-parentheses]
src/p_saveg.cpp:367:16: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
src/p_saveg.cpp:402:60: warning: comparison of integers of different signs: 'int' and 'unsigned int' [-Wsign-compare]
src/p_setup.cpp:1553:39: warning: format specifies type 'ptrdiff_t' (aka 'long') but the argument has type 'int' [-Wformat]
src/scripting/zscript/zcc_compile.cpp:293:74: warning: field 'AST' will be initialized after field 'mVersion' [-Wreorder]
src/swrenderer/drawers/r_thread.cpp:113:21: warning: comparison of integers of different signs: 'int' and 'size_t' (aka 'unsigned long') [-Wsign-compare]
- allow treatment as one-character string constants as character constants. This became necessary because name constants already use single quotes and are much harder to repurpose due to a higher degree of ambiguity.
- fixed: protected methods in structs were not usable.
It looks like the memory management at use here is not capable of maintaining multiple instances simultaneously and the camera textures create another scene drawer so the initialization of the main scene drawer has to be delayed until after the camera textures are done.
- replaced TStaticArray with regular TArrays.
They had incomplete implementations preventing proper cleanup of the level loading code. It makes more sense to add the missing methods to the regular TArray and use that.
This also makes some changes to how the game nodes are used to avoid creating a copy: If the head node's pointer is stored in a separate variable, no code needs to check which of the two arrays gets used.
src/p_glnodes.cpp:670:3: error: cannot jump from this goto statement to its label
src/p_glnodes.cpp:682:4: error: cannot jump from this goto statement to its label
- consolidated the code to calculate a sprite's display angle for all 3 renderers.
As it turned out, they all differed in their feature support because they had always been updated independently by different people.
The values were still 8 and 32 respectively which applied to hardware from last decade, but for modern mods these are simply too low. New values are 64 as minimum and 128 as default.
- added script access to a sector's colormap and specialcolors fields. (Writing only through dedicated functions because these fields are render state which may need to trigger some form of refresh if the renderer changes.)
With no 3D floors this appears to be ok, but there are so many places where colormaps are being set in the software renderer that I cannot guarantee that I got all of them correct. This will need some testing.
- moved testcolor and test fades into SWRenderer files.
These CCMDs work by hacking the default colormap and were never implemented for hardware rendering because they require many checks throughout the code.
This has increasingly become an obstacle with the hardware renderer, so now the values are being stored as plain data in the sector, with the software renderer getting the actual color tables when needed. While this is a bit slower than storing the pregenerated colormap, in realistic situations the added time is mostly negligible in the microseconds range.
This is only the parsing part, the arrays are not yet getting evaluated.
This required quite a hacky workaround because the gramma couldn't be made to accept the rule. The scanner will check if a 'static' token is immediately followed by a 'const' token and will combine both to a new 'staticconst' token that does not create conflicts with other rules.
- added a StartSlideshow ACS and ZScript command and extended the functionality to specify the slideshow's name when starting it.
This is for triggering any kind of intermission definition in the middle of a level - keep in mind that this may not be set up to loop!
This method was chosen because it avoids adding variable declarations to the global namespace which would have required a lot more work while polluting the grammar.
This way the global variables can be handled by a small bit of special coding in the struct generator.
If we have to write compiler specific code for micro-optimizations I am out.
The Posix compatible version nullified most the advantage on MSVC by writing out the XMM register to memory and then reading back the float.
That's not worth the hassle for an optimization that brings a few microseconds at best.
* dynamic lights also work in the true color software renderer and have been moved out of the OpenGL menu.
* created a separate software renderer menu and moved all relevant options there.
* delete non-applicable options when running in legacy mode.
* moved the OpenGL preferences menu one level up to eliminate a two-entry GL top level menu.
Removing this made me realize that calling the renderers' FakeFlat functions from the automap is inherently unsafe with the recent refactorings because there is absolutely no guarantee that the data may actually still be defined when the automap is being drawn.
So the best approach here is to give the automap its own FakeFlat function that runs independently of render data and assumptions of data preservation. This one can also be a lot simpler because it only needs the floor, not the ceiling info.
- optimized the math to get a plane equation from a linedef. The original code used a generic algorithm that knew nothing about the fact that Doom walls are always perfectly vertical. With this knowledge the plane calculation can be reduced to a lot less code because retrieving the normal is trivial in this special case.
- use the SSE2 rsqrtss instruction to calculate a wall's length, because this is by far the most frequent use of square roots in the GL renderer. So far this is only active on x64, it may be activated on 32 bit later as well, but only after it has been decided if 32 bit builds should be x87 or SSE2.
# Conflicts:
# src/gl/dynlights/gl_dynlight.cpp
# Conflicts:
# src/g_shared/a_dynlightdata.cpp
Since the true color software renderer also handles them there is no point keeping them on the GL side.
This also optimized how they are stored, because we no longer need to be aware of a base engine which doesn't have them.
- removed the LastCamera logic in RenderView. This code predates the first GZDoom release and apparently was only added because back then R_SetupFrame was not fully compatible with the hardware renderer. Today it is not needed anymore.
(Is there anyway to tone down GCC's warning level? It outputs too many false positives for potentially uninitialized variables in which the genuine errors get drowned.)
- enable precompiled headers for all non-system-specific MIDI devices.
- moved the native Windows and Mac MIDI devices into their respective sections in the project file so that they won't get compiled on the other ones.
This one was particularly nasty because Windows also defines a DWORD, but in Windows it is an unsigned long, not an unsigned int so changing types caused type conflicts and not all could be removed.
Those referring to the Windows type have to be kept, fortunately they are mostly in the Win32 directory, with a handful of exceptions elsewhere.
A major part of this device's implementation details about how to handle the callback were not encapsulated by the device class at all, they were #ifdef'd into the streamer class.
This puts everything into the device class which now exposes a clean interface to the rest of the game with no special handling aside from calling two additional virtual functions that are empty for the other devices
* make the critical section local to the respective platform instead of polluting everything with system specific symbols.
* moved system specific class declarations into the source file instead of having them in the global header.
This commit temporarily disables the Windows system device because it cannot be done without polluting the global header and still needs a bit of refactoring.
The approach being used here caused the entire sound system to be infested by windows.h, just to avoid copying around a handful of variables in one place, effectively preventing any compiling optimization.
Windows will now use the same internally defined structure for all MIDI processing which only for actual submission to the system player will be converted to the internal format.
Both files can now be included independently without causing problems.
This also required moving some inline functions into separate files and splitting off the GC definitions from dobject.h to ensure that r_defs does not need to pull in any part of the object hierarchy.
Most of those which still rely on ZDoom's own definition should be gone, unfortunately the code in files that include Windows headers is a gigantic mess with DWORDs being longs there intead of ints, so this needs to be done with care. DWORD should only remain where the Windows type is actually wanted.
- block creation of actors with the 'new' instruction. Unlike the above these cannot be made abstract because without ConstructNative they cannot be serialized.
This is one of two places that unconditionally pulled in all Windows headers into the GL code.
We also do not need the cruft for defining the standard integer types. GZDoom is C++11 which means that stdint.h will be present. So the madness with the definitions should be avoided to ensure that the types are always the same.
The code for constants was not optimal because the first operand of the instructions cannot be constant. This was solved by swapping it with the second choice which will always be non-constant.
The code for local variables did not allocate a new destination register and would overwrite the first parameter's variable.
Continuous integration is a cool thing: I completely forgot about addition of these frameworks because of my build environment which relies on static libraries and custom command line options
This addresses the main issue with TObjPtr, namely that using it required pulling in the entire class hierarchy in basic headers like r_defs which polluted nearly every single source file in the project.
It now uses a dedicated opcode instead of piggybacking on OP_CALL and it passes data that is closer to the VM. Symbols should be avoided at this level.
It also will skip the scope instruction if the code generator detects that both calling function and the self pointer type have the same scope, this assumes that subclasses cannot flip between UI and Play.
They are not needed for OP_NEW_K which can evaluate the class relations at compile time and for OP_NEW the calling function can also be checked at compile time, passing only the scope value itself.
According to MinGW headers (which is used for compiling libmpg123) ssize_t type must be 64-bit in size on 64-bit Windows
Moreover it was impossible to build GZDoom with the recent versions of libmpg123 because of wrong type redifinition
Tested MP3 music on Deus Vult II with Release x64 and OpenAL backend
Note that this completely disables the newly added keywords 'play' and 'ui' for unversioned code to allow using them as identifiers as I have found at least one mod that uses a variable named 'play' that would have been rendered broken otherwise.
This also disables many ZScript only keywords for other parsing jobs.
- made the self pointer of const functions readonly.
This seems to work fine. Both calling a non-const function and trying to assign a value to a member fail with an error message.
Clearscope is a dangerous context and should be limited to the minimum extent possible and preferably be blocked in user code.
This may still need some work on const functions but better have it in now.
This is to block modification of the planes directly. For future-proofness with renderer changes everything that alters these values should go through he function interface.
src/scripting/zscript/zcc_compile.cpp: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
src/p_acs.cpp: warning: format specifies type 'char *' but the argument has type 'PClass *' [-Wformat]
src/menu/menu.cpp:806:20: error: cannot pass non-trivial object of type 'FString' to variadic function; expected type from format string was 'char *' [-Wnon-pod-varargs]
