SVN handles and Git names are merged into one person.
Fix "undefined reference to `sampletimer()'" linking error using clang for a Win32 release build.
Fix errors and warnings in nedmalloc.h using clang for a Win32 release build.
Fix warnings involving printf() requiring explicit casts on tracked types.
Revert "compat.h: slightly reformat the NOWARN-related preprocessor directives."
Commit breaks MSVC builds by failing to provide NOWARN macros to MSVC.
Fix r3221: move the nedmalloc.h include to winbits.c.
Fix ebacktrace1.dll build with MinGW-GCC 4.7.2.
Remove many redundant (int16_t) casts and one now incorrect one.
The redundant ones are in code like this: s->cstat = (int16_t)32768; Because the value is eventually converted to the type of "s->cstat", any casts to integral types having at least as many bits are no-ops, signedness being irrelevant due to (probably any two's complement arch targeting compiler's) bit-pattern preserving semantics of these conversions. The now incorrect one is: if (lotag == (int16_t) 65535), "int32_t lotag" being read from a wall or sprite struct directly earlier. Now, with these members being unsigned, and (int16_t)65535 equalling -1, the check always fails. The correction fixes normal switches having such a lotag ending the level immediately. In short: integer casts before assignments are unnecessary, those in reads highly relevant!
Correct arithmetic comparisons of lo/hitags in actors.c.
Rewriting them in the obvious way, i.e. by casting the expression to int16_t first. (That is, this commit is the reverse of r3174, but with casts applied.) This fixes at least one regression: a FIREEXT with a hitag of 0 should not be linked with same- (that is, zero-) tagged SEENINEs or OOZFILTERs. Mind the corner cases!
Fix bound keys' commands being truncated in settings.cfg, control* cleanup.
- provide functions instead of messing with CONTROL_*Binds directly - comment out a few more unused functions - make clear what memory (alloc'd or const char *) 'keybind' members use - for keys with no name, use "<?>"
Fix a keyboard scancode oversight I introduced with my changes last night, apparently caused crashes
fix cycling picnum in 3D mode.
Using non-keypad [-]/[+] or LMB + mousewheel, it was broken when a void tile was in between.
Probably fix the zoom crash encountered with Mapster32 and CBP8.
Fix clipboard tile display in 3D mode.
demo.c: time G_DoMoveThings() using a double instead of an int32_t.
I think this may fix some negative "non-profiled overhead" that I have been seeing.
Address crash when model skins have invalid file names. Affected surfaces will not be drawn, but FPS will plummet as mdloadskin() is called repeatedly. A more optimal solution would either flag the surface as invalid (so the renderers skip it) or reject the entire model when defs are loaded.
Tools updates:
- Eliminate use of the C++ standard library from arttool. I improved some of the program's workings but it continues to corrupt my art files in certain circumstances. (Note: Not a regression; it happens with previous revisions.) - Construct "make utils" for the game side following the pattern of the engine. - Move ivfrate to this new distinction. - Minor Makefile cleanup and tweaks, mainly to facilitate CLANG=1 and to formalize text output in special cases.
Fix CPLUSPLUS=1 CLANG=1 building, and fix the "gotpic used attribute ignored" warning produced by clang.
MinGW C build fix
Fix the warnings when building with C++, add MSVC C++ build support. This also changes the internal type of lotags, hitags and cstat type fields to uint16_t instead of int16_t to clean up some dubious behavior wherein the code was using a value of 32768 as a flag in these fields for certain types of things, like invisible sprites (using the value as if it was uint16_t) where it was elsewhere checking if the value was < 0 (using the value as if it was int16_t). This change may break a few specific effects if any part of the relevant code was missed when looking for areas that needed to be addressed.
I think there's also a fix for the CON precache system breakage in here (lost it in my local tree when I started getting the C++ build working in MSVC, sorry!)
Fix r3155 fuckup :)
Fix useractor enemystayput breakage
Fix demo playback with diff application enabled.
This requires ripping out the .proj member of tiledata_t back into ProjectileData[] again because it was the only one in g_tile[] that changed during the course of a game. The g_tile[] array is now also correctly flagged as "constant throughout a game" (DS_NOCHK). Savegame version bumped.
Fix viewing through moving cameras.
It's still not pretty though since there's no interpolation from G_DrawRooms().
fix rotatesprite with no screen scaling and non-widescreen aspects.
Also, enable showing sprites in Mapster32's overhead view again, since that was one (the only?) use of that code path. The uninitialized read was introduced in r2927.
When found a camera sprite for viewscreen, set position etc. immediately.
Instead of later in P_ProcessInput(). This fixes one drawrooms() call in between where the sectnum is the camera's, but the position the player's, which can lead to a slow "outer view" render in Polymer.
Fix warning in jaudiolib due to void ptr arith.
Fix warnings from pointer math with the trackers.
Add automatic tracking to the sector[], wall[], sprite[] and tsprite[]
arrays; any write access to them will run the corresponding hook and write to the [sector/wall/sprite/tsprite]clean array. Note: tsprite and sprite use the same hook and require running a few more instructions per access in order to disambiguiate; this could be made more optimal (like the other arrays) by clearly separating the types in the game code. Note #2: taking a member's address currently marks it dirty because of tons of helper functions across the editor code. I don't know how many read-only accesses we have after taking a member address, but it could also be fixed with some finessing of the code.
Fix the C++ build on Linux with LTO.
Fix the Linux mapster32 build with LTO enabled.
