Commit graph

10380 commits

Author SHA1 Message Date
Bill Currie
0e40366c7f [ruamoko] Implement class conformsToProtocol
Seems to work nicely, too :)
2020-03-02 15:21:10 +09:00
Bill Currie
a906efac47 [ruamoko] Register protocols and their selectors 2020-03-02 15:20:34 +09:00
Bill Currie
5893bd7501 [qfcc] Catch erroneous negative builtin numbers
Setting a builtin number negative makes it a non-builtin function, but
possibly in the middle of another function. Not good.
2020-03-02 13:47:46 +09:00
Bill Currie
0db617719e [qfcc] Improve error messages for bad qc builtins
While global quakec functions could not be initialized to another
function, the error messages were rather obscure.
2020-03-02 13:47:46 +09:00
Bill Currie
78f552aa87 [ruamoko] Initialize static instances
This is one step closer to implementing conformsToProtocol. However,
protocols are not yet initialized correctly: they are not registered,
nor are their selectors.

While the static initializer list pointer was not written previously,
the module struct always came immediately after the symbols struct, and
the module version has so far always been 0. Thus, the list pointer is
correctly 0 for older progs and there's no need for a version bump.
2020-03-02 13:45:47 +09:00
Bill Currie
bc4bea79cb [ruamoko] Remove a stale FIXME
This was fixed when pr_def_t was introduced. I missed the comment at the
time.
2020-03-02 13:00:33 +09:00
Bill Currie
9ccff74fcf [qfcc] Emit only one instance per protocol per module
This is actually a double issue: when a class implementing a protocol
used the protocol in @protocol(), not only would the protocol get
emitted as part of the class data specifying that the class conforms to
the protocol, a second instance would be emitted again when @protocol()
was used. On top of that, only the instance referenced by @protocol()
would be initialized. Now, both class emission and @protocol() get their
protocol def from the same place and thus only one, properly
initialized, protocol instance is emitted.
2020-03-02 10:55:46 +09:00
Bill Currie
8021613b79 [qfcc] Fix missing protocol method lists
The problem was an erroneous assumption that the methods had to be
defined. Any class implementing a protocol must implement (and thus
define) the methods, but a protocol declaration cannot: it merely
declares the methods, and it's entirely possible for a module to see
only the protocol definition and not any classes implementing the
protocol.
2020-03-02 10:52:09 +09:00
Bill Currie
b6b7f9675f [qfcc] Emit static instance lists
For now, only protocols are in the list (gcc adds only static string
objects and qfcc doesn't do those yet, so not so far behind). qfprogs
dumps them.
2020-03-02 10:48:51 +09:00
Bill Currie
544d7de1ec [qfcc] Implement @protocol(foo)
Unlike gcc, qfcc requires foo to be defined, not just declared (I
suspect this is a bug in gcc, or even the ObjC spec), because allowing
forward declarations causes an empty (no methods) protocol to be
emitted, and then when the protocol is actually defined, one with
methods, resulting in two different versions of the same protocol, which
comments in the gnu objc runtime specifically state is a problem but is
not checked because it "never happens in practice" (found while
investigating gcc's behavior with @protocol and just what some of the
comments about static instance lists meant).
2020-03-02 10:42:26 +09:00
Bill Currie
254bf29bd4 [qfcc] Handle protocol forward declarations 2020-03-01 19:37:40 +09:00
Bill Currie
92cb3a5285 [qwaq] Flatten the hierarchy and do some cleanup 2020-03-01 18:25:02 +09:00
Bill Currie
80d9401eee [qfcc] Report errors for objects in function decls
The number of time's I've forgotten the * in a declaration in objective
code (probably thanks to C#'s lack of them).
2020-03-01 17:44:13 +09:00
Bill Currie
b544321609 [qfcc] Catch deferences to incomplete types
Reporting an error is so much more helpful than segmentation fault.
2020-03-01 17:43:28 +09:00
Bill Currie
277c64a460 [qfcc] Correct a typo 2020-03-01 17:28:51 +09:00
Bill Currie
4e4b553cc3 [gamecode] Take care of missed type alias stuff 2020-03-01 16:20:23 +09:00
Bill Currie
4d076da4b6 [qwaq] Fix warning hidden by type alias bug 2020-03-01 16:15:33 +09:00
Bill Currie
caa297b756 [qfcc] Remove type alias encoding
It proved to be too fragile in its current implementation. It broke
pointers to incomplete structs and switch enum checking, and getting it
to work for other things was overly invasive. I still want the encoding,
but need to come up with something more robust.a
2020-03-01 16:13:18 +09:00
Bill Currie
1033716b2b [qfcc] Fix some curly space 2020-03-01 13:53:18 +09:00
Bill Currie
df2ed4b086 [qwaq] Move from tools to ruamoko
This fixes the dependency issues between qwaq and ruamoko. qwaq is
actually older than ruamoko. That little language feature test has come
a long way.

However, I'm considering moving to non-recursive make, but...
2020-03-01 00:55:15 +09:00
Bill Currie
c079eb851b [qwaq] Start work on the actual app
It doesn't look good, but it does have panel based windows working, and
using objects. Won't build reliably right now due to qwaq being in tools
and thus building before ruamoko, but I'll fix that next.
2020-03-01 00:40:55 +09:00
Bill Currie
271d836cd2 [qfcc] Catch static class instances in structs 2020-02-29 21:09:24 +09:00
Bill Currie
e93ca9d828 [qfcc] Fix infinite loop in linker 2020-02-29 20:13:25 +09:00
Bill Currie
8b6d05a4dd [qwaq] Ensure mouse motion reporting gets turned off
It seems that xterm automatically disables it when ncurses shuts down and
mate-terminal does not, or maybe a different version of something. Still,
good to clean up properly.
2020-02-29 16:36:23 +09:00
Bill Currie
75f4cebf0e Clean up after removal of uint32.h
I guess I forgot to check for its usage in header files, and the fact that
it didn't get cleaned out from being installed hid the error.
2020-02-29 16:11:07 +09:00
Bill Currie
ae532870c4 [qwaq] Implement basic color support 2020-02-29 14:48:18 +09:00
Bill Currie
ec4e9b326d [qwaq] Don't call wrefresh in cmd_mvwaddstr
There is now an implementation for wrefresh.
2020-02-29 14:43:08 +09:00
Bill Currie
f5f50ae231 [qwaq] Make stdscr available to progs 2020-02-29 14:38:54 +09:00
Bill Currie
5e4677f8d5 [qwaq] Implement the immediately useful panel functions 2020-02-29 13:06:58 +09:00
Bill Currie
17005637ca [qwaq] Rename the internal function names
Now they reflect the curses functions they wrap. The externally visible
builtin names are not changed because the parameters are in x, y order
rather than curses' y, x order.
2020-02-29 12:52:13 +09:00
Bill Currie
ea69921e6a [qwaq] Validate window before acquiring string
If the window is invalid and recovery is done, string ids will leak if
acquired before validation.

Afterwards, make the rest of the builtin wrappers consistent: extract
parameters, validate, [acquire resources], generate command.
2020-02-29 12:33:45 +09:00
Bill Currie
0119660b01 [qwaq] Convert remaining functions to command queue
Now that the initial prototype seems to be working well, it's time to
implement more commands. I might have to do some wrappers for actual
command writing (and result reading) as it looks like there will be a
lot of nearly identical code.
2020-02-29 11:44:43 +09:00
Bill Currie
513c808875 [qwaq] Make some bad ascii art 2020-02-29 02:07:47 +09:00
Bill Currie
bd98d1d9fb [qwaq] Prepare for threading
So far, no threading has been set up, and only window creation and
printing have been updated, but the basics of the design seem to be
sound.

The builtin functions now no longer call ncurses directly: the build
commands and write them to a command buffer.

Commands that have return values (eg, window creation) write their
results to a results buffer that the originating builtin function
reads. Builtin functions that expect a result "poll" the results buffer
for the correct result (marked by the same command). In a single
UI-thread environment, the results should always be in the same order as
the commands, and in a multi-UI-thread environment, things should
(fingers crossed) sort themselves out as ONE of the threads will be the
originator of the next available result.

Strings in commands (eg, for printing) are handled by acquiring a string
id (index into an array of dstring_t) and including the string id in the
written command. The string id is released upon completion of the
command.

Builtin functions write commands, acquire string ids, and read results.

The command processor reads commands, releases string ids, and writes
results.

Since commands, string ids, and results are all in ring buffers, and
assuming there is only one thread running the builtin functions and only
one thread processing commands (there can be only one because ncurses is
not thread-safe), then there should never be any contention on the
buffers. Of course, if there are multiple threads running the builtin
functions, then locking will be required on the builtin function side.
2020-02-29 01:45:33 +09:00
Bill Currie
644ef93dde [qwaq] Create some ring-buffer macros
I expect I will need several messaging buffers, and ring buffers tend to
be quite robust. Replacing the event buffer code with the macros made
testing easy.
2020-02-28 22:27:29 +09:00
Bill Currie
6a58dcdddd [qwaq] Fix lost output
Turns out all I needed was a refresh() after initialization.
2020-02-27 21:38:55 +09:00
Bill Currie
dd25bf5dfe [qwaq] Read mouse movements
Many thanks to https://gist.github.com/sylt/93d3f7b77e7f3a881603 for the
necessary escape sequence to get xterm reporting mouse movement events.
2020-02-27 21:22:10 +09:00
Bill Currie
a3ed5926b9 [qwaq] Remove unnecessary fields from mouse events
id and z seem to always be 0.

Ironically, it turns out that the work needed for "int id" and "large"
struct nil init wasn't strictly necessary to get to this point, but
without having done that work, I wouldn't know :)
2020-02-27 21:08:12 +09:00
Bill Currie
08bf8a04e4 [qwaq] Implement an event queue
It seems to have an issue with a bogus clearing of the screen, but the
basics seem to be working.
2020-02-27 21:07:56 +09:00
Bill Currie
b4aebc120e [qfcc] Treat { } as nil for initializing compound types 2020-02-27 20:30:39 +09:00
Bill Currie
7c9072aebf [qfcc] Create struct fields for "type typename"
Such declarations were being lost, thus in the following, the id field
never got added:
typedef struct qwaq_mevent_s {
    int         id;
    int         x, y, z;
    int         buttons;
} qwaq_mevent_t;
2020-02-27 17:50:11 +09:00
Bill Currie
dbbb8a1396 [qfcc] Fix syntax error for id as a field name
event.e.mouse.id produced a syntax error, which is contrary to
Objective-C.
2020-02-27 17:43:39 +09:00
Bill Currie
e8c357393f [qwaq] Clear qwaq's print buffer
Didn't realized PR_Sprintf appended. Or, more likely, I had forgotten
because I imagine Deek and I discussed it at the time.
2020-02-27 02:11:54 +09:00
Bill Currie
789f263855 Use get_strref() correctly
I had forgotten that it works only for dynamic strings.
2020-02-27 02:07:53 +09:00
Bill Currie
c9186c8296 Clear the statement line buffer
The problem with moving to a cached buffer is it needs to be cleared.
Fixes the weird output of qfprogs and tracing (not missing local defs,
though).
2020-02-27 01:25:41 +09:00
Bill Currie
edde4bad15 Create a basic hello world
And it has begun. It has some problems, but worse, it seems I broke
qfprogs and maybe pr_debug.c.
2020-02-27 01:18:38 +09:00
Bill Currie
126f8502bd Start working on a qwaq console tool
The intention is it will hopefully become a debugger. It will certainly
help with development of the progs engine.
2020-02-26 22:10:59 +09:00
Bill Currie
62b541793d [gamecode] Remove all use of va
With this, the VA is very close to being safe to use in a threaded
environment (so long as each VM is used by only one thread). Just the
debug file hash and source paths to sort out.
2020-02-26 19:30:10 +09:00
Bill Currie
ddd007e2d5 [dstring] Return the string instead of printed size
Other than consistency with printf(), I'm not sure why we went with the
printed size as the return value; returning the resultant strings makes
much more sense as dsprintf() (etc) can then be used as a safe va()
2020-02-26 19:00:19 +09:00
Bill Currie
36bc139b27 Avoid use of va in PR_Sprintf
With this, progs strings are thread safe so long as only one thread
tries to use the VM.
2020-02-26 18:35:19 +09:00