File di log originale · 20191019-8229.txt
------------------------------------------------------------------------ r8229 | terminx | 2019-10-19 17:18:16 -0700 (Sat, 19 Oct 2019) | 1 line I thought this would work but I'm stupid :D ------------------------------------------------------------------------ r8228 | terminx | 2019-10-19 16:49:02 -0700 (Sat, 19 Oct 2019) | 1 line Tweak the priority classes set by win_priorityclass ------------------------------------------------------------------------ r8227 | terminx | 2019-10-19 16:48:59 -0700 (Sat, 19 Oct 2019) | 1 line Fix issue causing maps to sometimes fail to load ------------------------------------------------------------------------ r8226 | terminx | 2019-10-19 16:48:55 -0700 (Sat, 19 Oct 2019) | 1 line Band-aid fix an issue with the tile packer preventing MSVC builds from working ------------------------------------------------------------------------ r8225 | terminx | 2019-10-19 16:48:51 -0700 (Sat, 19 Oct 2019) | 1 line Prefer uint32_t over unsigned int ------------------------------------------------------------------------ r8224 | terminx | 2019-10-19 16:48:44 -0700 (Sat, 19 Oct 2019) | 1 line The constening ------------------------------------------------------------------------ r8223 | terminx | 2019-10-19 16:48:40 -0700 (Sat, 19 Oct 2019) | 1 line Add mus_adlibstereo cvar to enable and disable AdLib stereo mode ------------------------------------------------------------------------ r8222 | terminx | 2019-10-19 16:48:35 -0700 (Sat, 19 Oct 2019) | 1 line Add a cvar array for audiolib ------------------------------------------------------------------------ r8221 | terminx | 2019-10-19 16:48:31 -0700 (Sat, 19 Oct 2019) | 1 line Further small midi cleanup changes--nothing functional ------------------------------------------------------------------------ r8220 | terminx | 2019-10-19 16:48:26 -0700 (Sat, 19 Oct 2019) | 1 line Further AdLib cleanup... leaving this in a separate commit in case I fucked something up ------------------------------------------------------------------------ r8219 | terminx | 2019-10-19 16:48:20 -0700 (Sat, 19 Oct 2019) | 1 line Further audiolib work... I think this is almost done! ------------------------------------------------------------------------ r8218 | terminx | 2019-10-19 16:48:08 -0700 (Sat, 19 Oct 2019) | 1 line More audiolib work ------------------------------------------------------------------------ r8217 | terminx | 2019-10-19 16:47:59 -0700 (Sat, 19 Oct 2019) | 3 lines Sort lists of things in the makefile I'm not gonna lie, I got some kind of weird OCD pleasure out of this. ------------------------------------------------------------------------ r8216 | terminx | 2019-10-19 16:47:54 -0700 (Sat, 19 Oct 2019) | 3 lines Audiolib rework WIP This attempts to rectify the differences between versions of JFAudiolib created after we forked the code, and the extra features contained in Nuke.YKT's fork of our version. ------------------------------------------------------------------------ r8215 | terminx | 2019-10-19 16:47:42 -0700 (Sat, 19 Oct 2019) | 3 lines New files for audiolib This is in a separate commit just to make the changes to existing files easier to go through. ------------------------------------------------------------------------ r8214 | terminx | 2019-10-19 16:47:29 -0700 (Sat, 19 Oct 2019) | 3 lines Use SDL audiolib driver on Windows This shitcans the SDL_mixer requirement but leaves platforms other than Windows without built-in MIDI playback capability until Nuke.YKT's OPL3 emulator is merged. This also reworks sdlmusic.cpp into music_external.cpp, including an untested port of the code to Windows. ------------------------------------------------------------------------ r8213 | terminx | 2019-10-19 16:47:19 -0700 (Sat, 19 Oct 2019) | 1 line Misc audiolib cleanups ------------------------------------------------------------------------ r8212 | terminx | 2019-10-19 16:47:12 -0700 (Sat, 19 Oct 2019) | 1 line Add snd_tryformats cvar to disable automatic loading of .ogg and .flac versions of sound and music ------------------------------------------------------------------------ r8211 | terminx | 2019-10-19 16:47:08 -0700 (Sat, 19 Oct 2019) | 3 lines With MinGW GCC, you get a different __cpuid() depending on whether or not you include cpuid.h. That is strange. This fixes sysReadCPUID() to work on more than just Windows. ------------------------------------------------------------------------ r8210 | terminx | 2019-10-19 16:47:04 -0700 (Sat, 19 Oct 2019) | 1 line Add -Wno-deprecated-declarations to the warnings for GCC 9 and up ------------------------------------------------------------------------ r8209 | terminx | 2019-10-19 16:47:01 -0700 (Sat, 19 Oct 2019) | 1 line Remove Android.mk ------------------------------------------------------------------------ r8208 | terminx | 2019-10-19 16:46:58 -0700 (Sat, 19 Oct 2019) | 1 line Fix tiny header issue ------------------------------------------------------------------------ r8207 | terminx | 2019-10-19 16:46:54 -0700 (Sat, 19 Oct 2019) | 1 line Don't try videoSetGameMode() if there aren't any valid modes ------------------------------------------------------------------------ r8206 | terminx | 2019-10-19 16:46:50 -0700 (Sat, 19 Oct 2019) | 1 line Patch from Nuke.YKT to fix polygon leakage issue with NPOT skies in Polymost ------------------------------------------------------------------------ r8205 | terminx | 2019-10-19 16:46:46 -0700 (Sat, 19 Oct 2019) | 1 line Bump default cache1d size to 128MB in the editor and add support for the cachesize def token ------------------------------------------------------------------------ r8204 | terminx | 2019-10-19 16:46:41 -0700 (Sat, 19 Oct 2019) | 3 lines cache1d rework This turns cache1d into a reusable class that can be used to implement multiple caches for different types of data. Also adds a "cacheinfo" console command that prints out some statistics about the cache. ------------------------------------------------------------------------ r8203 | terminx | 2019-10-19 16:46:34 -0700 (Sat, 19 Oct 2019) | 1 line Move DEBUG_ALLOCACHE_AS_MALLOC crap to the bottom of cache1d.cpp instead of having preprocessor garbage in every function ------------------------------------------------------------------------ r8202 | terminx | 2019-10-19 16:46:30 -0700 (Sat, 19 Oct 2019) | 3 lines This seems like it should be allocated permanently (value 255) instead of merely marked as locked (value 200) There doesn't seem to be any functionality implemented to re-load voxels if they're removed from the cache, and blocks marked as locked can still be pushed out of the cache if the cache is full and a large allocation won't fit anywhere else. ------------------------------------------------------------------------ r8201 | terminx | 2019-10-19 16:46:26 -0700 (Sat, 19 Oct 2019) | 3 lines Better cache1d enums plus some general cache1d maintenance Renamed variables, added usage of said enums, etc. ------------------------------------------------------------------------ r8200 | terminx | 2019-10-19 16:46:17 -0700 (Sat, 19 Oct 2019) | 1 line Speed up G_ReadGLFrame() by just grabbing a chunk of memory from cache1d instead of doing a huge malloc() ------------------------------------------------------------------------ r8199 | terminx | 2019-10-19 16:46:12 -0700 (Sat, 19 Oct 2019) | 3 lines Separate the filesystem portion of cache1d.cpp and move it into vfs.cpp Yes, the engine tools (kextract, etc) still compile. ------------------------------------------------------------------------ r8198 | terminx | 2019-10-19 16:46:04 -0700 (Sat, 19 Oct 2019) | 1 line Fix misplaced files in VS solution filters ------------------------------------------------------------------------ r8197 | terminx | 2019-10-19 16:46:00 -0700 (Sat, 19 Oct 2019) | 3 lines audiolib: pull memory for vorbis_data struct from cache1d This gets rid of one calloc() and free() per vorbis voice playback, and most of the time allows reuse of the block from cache1d across multiple uses of the VoiceNode. ------------------------------------------------------------------------ r8196 | terminx | 2019-10-19 16:45:56 -0700 (Sat, 19 Oct 2019) | 1 line Add CACHE1D_ENTRY_FREE and CACHE1D_ENTRY_PERMANENT enums ------------------------------------------------------------------------ r8195 | terminx | 2019-10-19 16:45:49 -0700 (Sat, 19 Oct 2019) | 3 lines Wait to run G_MoveLoop() until a frame has just been rendered This should give G_MoveLoop() a better chance to not run past the time when another frame is to be drawn. ------------------------------------------------------------------------ r8194 | terminx | 2019-10-19 16:45:45 -0700 (Sat, 19 Oct 2019) | 1 line Silence a warning ------------------------------------------------------------------------ r8193 | terminx | 2019-10-19 16:45:41 -0700 (Sat, 19 Oct 2019) | 1 line Replace usages of deprecated LZ4_decompress_fast() API with LZ4_decompress_safe() as recommended by current LZ4 documentation ------------------------------------------------------------------------ r8192 | terminx | 2019-10-19 16:45:36 -0700 (Sat, 19 Oct 2019) | 1 line Add #pragma once to debugbreak.h and remove it from separate inclusion in a couple of translation units as we're including it directly in compat.h now ------------------------------------------------------------------------ r8191 | terminx | 2019-10-19 16:45:31 -0700 (Sat, 19 Oct 2019) | 3 lines Add fatal_exit(), a slightly different method of terminating the process upon error On Windows, this writes an event to the system event log for later retrieval. ------------------------------------------------------------------------ r8190 | terminx | 2019-10-19 16:45:24 -0700 (Sat, 19 Oct 2019) | 1 line Use EXIT_FAILURE and EXIT_SUCCESS from stdlib.h when calling exit() ------------------------------------------------------------------------ r8189 | terminx | 2019-10-19 16:45:18 -0700 (Sat, 19 Oct 2019) | 1 line Use zpl_virtual_memory_page_size() to implement our Bgetpagesize() ------------------------------------------------------------------------ r8188 | terminx | 2019-10-19 16:45:14 -0700 (Sat, 19 Oct 2019) | 1 line Make Bgetsysmemsize() return size_t ------------------------------------------------------------------------ r8187 | terminx | 2019-10-19 16:45:10 -0700 (Sat, 19 Oct 2019) | 1 line Use zpl_vm (backed by VirtualAlloc or mmap) for larger memory chunks ------------------------------------------------------------------------ r8186 | terminx | 2019-10-19 16:45:06 -0700 (Sat, 19 Oct 2019) | 1 line Update LZ4 to e8baeca51ef2003d6c9ec21c32f1563fef1065b9 ------------------------------------------------------------------------ r8185 | terminx | 2019-10-19 16:45:01 -0700 (Sat, 19 Oct 2019) | 1 line Update xxHash to 1ea98d6a38300f7224869de856a876a2050cdf78 ------------------------------------------------------------------------ r8184 | terminx | 2019-10-19 16:44:57 -0700 (Sat, 19 Oct 2019) | 1 line Add const to dest parameter of MV_Reverb() ------------------------------------------------------------------------ r8183 | terminx | 2019-10-19 16:44:53 -0700 (Sat, 19 Oct 2019) | 3 lines Use templates for audiolib mixing functions Patch from NY00123. ------------------------------------------------------------------------ r8182 | terminx | 2019-10-19 16:44:48 -0700 (Sat, 19 Oct 2019) | 3 lines This lets sys_timer mode 4 (rdtsc) work on ARM ...probably ------------------------------------------------------------------------ r8181 | terminx | 2019-10-19 16:44:45 -0700 (Sat, 19 Oct 2019) | 1 line Fix issue with crouch toggle and jetpack ------------------------------------------------------------------------ r8180 | terminx | 2019-10-19 16:44:41 -0700 (Sat, 19 Oct 2019) | 1 line Explicitly yield audio data fill thread when work is finished ------------------------------------------------------------------------ r8179 | terminx | 2019-10-19 16:44:36 -0700 (Sat, 19 Oct 2019) | 5 lines Add portions of the ZPL library, a public domain collection of useful helper functions and structures See https://github.com/zpl-c/zpl I'm not including all of it, but I have plans for some of what is there. ------------------------------------------------------------------------ r8178 | terminx | 2019-10-19 16:44:30 -0700 (Sat, 19 Oct 2019) | 3 lines Remove limit of MAXCACHEOBJECTS in favor of dynamic allocation/expansion of the cache index Somewhat based on a patch from Doom64hunter. ------------------------------------------------------------------------ r8177 | terminx | 2019-10-19 16:44:26 -0700 (Sat, 19 Oct 2019) | 1 line Make the lock ptr the first member of the cactype struct as it's the most accessed ------------------------------------------------------------------------ r8176 | terminx | 2019-10-19 16:44:22 -0700 (Sat, 19 Oct 2019) | 1 line These changes are on top of the enet.h from the previous commit and are what I needed to build under MSYS2 without warnings or errors ------------------------------------------------------------------------ r8175 | terminx | 2019-10-19 16:44:15 -0700 (Sat, 19 Oct 2019) | 1 line Swap out our old deprecated version of ENet for a single-file header-based fork, including features like IPv6 support ------------------------------------------------------------------------ r8174 | terminx | 2019-10-19 16:44:03 -0700 (Sat, 19 Oct 2019) | 7 lines Significant reworks to everything in winbits.cpp Important cvars after this commit: sys_timer win_systemtimermode win_priorityclass ------------------------------------------------------------------------ r8173 | terminx | 2019-10-19 16:43:55 -0700 (Sat, 19 Oct 2019) | 1 line Add missing header to VS project ------------------------------------------------------------------------ r8172 | terminx | 2019-10-19 16:43:51 -0700 (Sat, 19 Oct 2019) | 1 line Add cvar to control ntdll.dll timer voodoo ------------------------------------------------------------------------ r8171 | terminx | 2019-10-19 16:43:46 -0700 (Sat, 19 Oct 2019) | 1 line Implement sys_timer cvar with several different choices of timers to control frame pacing ------------------------------------------------------------------------ r8170 | terminx | 2019-10-19 16:43:38 -0700 (Sat, 19 Oct 2019) | 1 line This is useless ------------------------------------------------------------------------ r8169 | terminx | 2019-10-19 16:43:35 -0700 (Sat, 19 Oct 2019) | 1 line Rename timerUpdate() and timerGetRate() to timerUpdateClock() and timerGetClockRate(), because they both only apply to totalclock ------------------------------------------------------------------------ r8168 | terminx | 2019-10-19 16:43:30 -0700 (Sat, 19 Oct 2019) | 1 line Use nanosleep instead of usleep ------------------------------------------------------------------------ r8167 | terminx | 2019-10-19 16:43:26 -0700 (Sat, 19 Oct 2019) | 1 line Patch from Doom64hunter to address a couple of issues with per-map art ------------------------------------------------------------------------ r8166 | terminx | 2019-10-19 16:43:22 -0700 (Sat, 19 Oct 2019) | 1 line Renderer sprite sorting patch from Nuke.YKT ------------------------------------------------------------------------ r8165 | terminx | 2019-10-19 16:43:18 -0700 (Sat, 19 Oct 2019) | 1 line Explicitly implement timerGetTicksU64() with chrono::high_resolution_timer and use chrono::steady_clock for regular timerGetHiTicks() and totalclock ------------------------------------------------------------------------ r8164 | terminx | 2019-10-19 16:43:15 -0700 (Sat, 19 Oct 2019) | 3 lines Recalculate frame limiter delay every time a frame is drawn This is an attempt to work around issues that arise when timerGetTicksU64() isn't monotonic. Timing situations like that shouldn't happen, but some systems have pretty messed up configurations. ------------------------------------------------------------------------ r8163 | terminx | 2019-10-19 16:43:11 -0700 (Sat, 19 Oct 2019) | 1 line Let's try something different for r_swapcomplete ------------------------------------------------------------------------ r8162 | terminx | 2019-10-19 16:43:07 -0700 (Sat, 19 Oct 2019) | 1 line Remove obsolete win_uninit() function ------------------------------------------------------------------------ r8161 | terminx | 2019-10-19 16:43:02 -0700 (Sat, 19 Oct 2019) | 1 line Disable SDL's built-in signal handler ------------------------------------------------------------------------ r8160 | terminx | 2019-10-19 16:42:59 -0700 (Sat, 19 Oct 2019) | 1 line Call debug_break() from attach_debugger_here() in builds with DEBUGGINGAIDS enabled ------------------------------------------------------------------------ r8159 | terminx | 2019-10-19 16:42:55 -0700 (Sat, 19 Oct 2019) | 1 line This old 3dfx and SiS hardware isn't capable of running our GL renderer and never will be ------------------------------------------------------------------------ r8158 | terminx | 2019-10-19 16:42:51 -0700 (Sat, 19 Oct 2019) | 1 line Equivalent to jfbuild commit b6a666745cea21c6f8193b046bcdd0636fb7c166 ------------------------------------------------------------------------ r8157 | terminx | 2019-10-19 16:42:47 -0700 (Sat, 19 Oct 2019) | 1 line Simplify SDL subsystem init a little ------------------------------------------------------------------------ r8156 | terminx | 2019-10-19 16:42:44 -0700 (Sat, 19 Oct 2019) | 3 lines Rename BUILD_xxx debug env vars to EDUKE32_xxx Add EDUKE32_NO_NVIDIA_THREADED_OPTIMIZATIONS for Linux and EDUKE32_NO_XINPUT for Windows ------------------------------------------------------------------------ r8155 | terminx | 2019-10-19 16:42:39 -0700 (Sat, 19 Oct 2019) | 1 line Don't send a mouse event when clicking to re-focus the game window ------------------------------------------------------------------------ r8154 | terminx | 2019-10-19 16:42:36 -0700 (Sat, 19 Oct 2019) | 1 line Fix oversight where the check for whether the loaded version of SDL is new enough was never updated for SDL2 ------------------------------------------------------------------------ r8153 | terminx | 2019-10-19 16:42:32 -0700 (Sat, 19 Oct 2019) | 1 line GLU isn't loaded with SDL, so calling SDL_GetError() here is pretty useless ------------------------------------------------------------------------ r8152 | terminx | 2019-10-19 16:42:28 -0700 (Sat, 19 Oct 2019) | 3 lines Use aligned allocation for the block gamecontrollerdb.txt is read into I know, unnecessary. ------------------------------------------------------------------------ r8151 | terminx | 2019-10-19 16:42:24 -0700 (Sat, 19 Oct 2019) | 1 line SDL_free() is not for pointers to SDL_RWops structures... ------------------------------------------------------------------------ r8150 | terminx | 2019-10-19 16:42:21 -0700 (Sat, 19 Oct 2019) | 1 line Don't use the SDL joystick API with SDL2, just the updated one for game controllers ------------------------------------------------------------------------ r8149 | terminx | 2019-10-19 16:42:17 -0700 (Sat, 19 Oct 2019) | 1 line Add optional duration parameter to idle() ------------------------------------------------------------------------ r8148 | terminx | 2019-10-19 16:42:13 -0700 (Sat, 19 Oct 2019) | 1 line Clamp nextPageDelay in G_FPSLimit() to 0.0 - g_frameDelay ------------------------------------------------------------------------ r8147 | terminx | 2019-10-19 16:42:10 -0700 (Sat, 19 Oct 2019) | 1 line Increase timer resolution on Windows ------------------------------------------------------------------------ r8146 | terminx | 2019-10-19 16:42:05 -0700 (Sat, 19 Oct 2019) | 3 lines Remove win_allowtaskswitching() We weren't ever calling win_allowtaskswitching(0) anywhere, and when I tried it, it did not work. Broken functionality to try and block alt + tab is not necessary. ------------------------------------------------------------------------ r8145 | terminx | 2019-10-19 16:42:01 -0700 (Sat, 19 Oct 2019) | 1 line Use timeGetTime() to implement timerGetTicks() on Windows ------------------------------------------------------------------------ r8144 | terminx | 2019-10-19 16:41:58 -0700 (Sat, 19 Oct 2019) | 1 line Remove dead function declarations ------------------------------------------------------------------------ r8143 | terminx | 2019-10-19 16:41:54 -0700 (Sat, 19 Oct 2019) | 1 line This function declaration was in the wrong header ------------------------------------------------------------------------ r8142 | terminx | 2019-10-19 16:41:50 -0700 (Sat, 19 Oct 2019) | 1 line Update xxHash to git commit cd0f5c22095c49c35104dca86620322801f14893 ------------------------------------------------------------------------ r8141 | terminx | 2019-10-19 16:41:44 -0700 (Sat, 19 Oct 2019) | 1 line Use timerGetHiTicks() instead of timerGetTicks() for timing in osdfunc_fileinfo() ------------------------------------------------------------------------ r8140 | terminx | 2019-10-19 16:41:40 -0700 (Sat, 19 Oct 2019) | 1 line G_HandleAsync() -> gameHandleEvents() ------------------------------------------------------------------------ r8139 | terminx | 2019-10-19 16:41:35 -0700 (Sat, 19 Oct 2019) | 3 lines Partial timer cleanup After this revision, the only place timerUpdate() is called is from within handleevents(). S_Cleanup(), MUSIC_Update(), and G_HandleSpecialKeys() are now called from a timer callback set with timerSetCallback(). This more or less deprecates the usage of faketimerhandler() in EDuke32 and Mapster32, but other games still rely on the functionality. ------------------------------------------------------------------------ r8138 | terminx | 2019-10-19 16:41:29 -0700 (Sat, 19 Oct 2019) | 1 line Reword stupid error message that has probably never even been seen ------------------------------------------------------------------------ r8137 | terminx | 2019-10-19 16:41:26 -0700 (Sat, 19 Oct 2019) | 3 lines ifdef DEBUGGINGAIDS for this quick exit crap this is more like regular AIDS than DEBUGGINGAIDS though ------------------------------------------------------------------------ r8136 | terminx | 2019-10-19 16:41:22 -0700 (Sat, 19 Oct 2019) | 1 line Use tabledivide64() in timerUpdate() ------------------------------------------------------------------------ r8135 | terminx | 2019-10-19 16:41:18 -0700 (Sat, 19 Oct 2019) | 1 line Ugh, apparently we rely on this for FURY ------------------------------------------------------------------------ See http://svn.eduke32.com/listing.php?repname=eduke32 for more details.
Tweak the priority classes set by win_priorityclass
Fix issue causing maps to sometimes fail to load
Band-aid fix an issue with the tile packer preventing MSVC builds from working
Prefer uint32_t over unsigned int
The constening
Add mus_adlibstereo cvar to enable and disable AdLib stereo mode
Add a cvar array for audiolib
Further small midi cleanup changes--nothing functional
Further AdLib cleanup... leaving this in a separate commit in case I fucked something up
Further audiolib work... I think this is almost done!
More audiolib work
Sort lists of things in the makefile
I'm not gonna lie, I got some kind of weird OCD pleasure out of this.
Audiolib rework WIP
This attempts to rectify the differences between versions of JFAudiolib created after we forked the code, and the extra features contained in Nuke.YKT's fork of our version.
New files for audiolib
This is in a separate commit just to make the changes to existing files easier to go through.
Use SDL audiolib driver on Windows
This shitcans the SDL_mixer requirement but leaves platforms other than Windows without built-in MIDI playback capability until Nuke.YKT's OPL3 emulator is merged. This also reworks sdlmusic.cpp into music_external.cpp, including an untested port of the code to Windows.
Misc audiolib cleanups
Add snd_tryformats cvar to disable automatic loading of .ogg and .flac versions of sound and music
With MinGW GCC, you get a different __cpuid() depending on whether or not you include cpuid.h. That is strange.
This fixes sysReadCPUID() to work on more than just Windows.
Add -Wno-deprecated-declarations to the warnings for GCC 9 and up
Remove Android.mk
Fix tiny header issue
Don't try videoSetGameMode() if there aren't any valid modes
Patch from Nuke.YKT to fix polygon leakage issue with NPOT skies in Polymost
Bump default cache1d size to 128MB in the editor and add support for the cachesize def token
cache1d rework
This turns cache1d into a reusable class that can be used to implement multiple caches for different types of data. Also adds a "cacheinfo" console command that prints out some statistics about the cache.
Move DEBUG_ALLOCACHE_AS_MALLOC crap to the bottom of cache1d.cpp instead of having preprocessor garbage in every function
This seems like it should be allocated permanently (value 255) instead of merely marked as locked (value 200)
There doesn't seem to be any functionality implemented to re-load voxels if they're removed from the cache, and blocks marked as locked can still be pushed out of the cache if the cache is full and a large allocation won't fit anywhere else.
Better cache1d enums plus some general cache1d maintenance
Renamed variables, added usage of said enums, etc.
Speed up G_ReadGLFrame() by just grabbing a chunk of memory from cache1d instead of doing a huge malloc()
Separate the filesystem portion of cache1d.cpp and move it into vfs.cpp
Yes, the engine tools (kextract, etc) still compile.
Fix misplaced files in VS solution filters
pull memory for vorbis_data struct from cache1d
This gets rid of one calloc() and free() per vorbis voice playback, and most of the time allows reuse of the block from cache1d across multiple uses of the VoiceNode.
Add CACHE1D_ENTRY_FREE and CACHE1D_ENTRY_PERMANENT enums
Wait to run G_MoveLoop() until a frame has just been rendered
This should give G_MoveLoop() a better chance to not run past the time when another frame is to be drawn.
Silence a warning
Replace usages of deprecated LZ4_decompress_fast() API with LZ4_decompress_safe() as recommended by current LZ4 documentation
Add #pragma once to debugbreak.h and remove it from separate inclusion in a couple of translation units as we're including it directly in compat.h now
Add fatal_exit(), a slightly different method of terminating the process upon error
On Windows, this writes an event to the system event log for later retrieval.
Use EXIT_FAILURE and EXIT_SUCCESS from stdlib.h when calling exit()
Use zpl_virtual_memory_page_size() to implement our Bgetpagesize()
Make Bgetsysmemsize() return size_t
Use zpl_vm (backed by VirtualAlloc or mmap) for larger memory chunks
Update LZ4 to e8baeca51ef2003d6c9ec21c32f1563fef1065b9
Update xxHash to 1ea98d6a38300f7224869de856a876a2050cdf78
Add const to dest parameter of MV_Reverb()
Use templates for audiolib mixing functions
Patch from NY00123.
This lets sys_timer mode 4 (rdtsc) work on ARM
...probably
Fix issue with crouch toggle and jetpack
Explicitly yield audio data fill thread when work is finished
Add portions of the ZPL library, a public domain collection of useful helper functions and structures
See https://github.com/zpl-c/zpl I'm not including all of it, but I have plans for some of what is there.
Remove limit of MAXCACHEOBJECTS in favor of dynamic allocation/expansion of the cache index
Somewhat based on a patch from Doom64hunter.
Make the lock ptr the first member of the cactype struct as it's the most accessed
These changes are on top of the enet.h from the previous commit and are what I needed to build under MSYS2 without warnings or errors
Swap out our old deprecated version of ENet for a single-file header-based fork, including features like IPv6 support
Significant reworks to everything in winbits.cpp
Important cvars after this commit: sys_timer win_systemtimermode win_priorityclass
Add missing header to VS project
Add cvar to control ntdll.dll timer voodoo
Implement sys_timer cvar with several different choices of timers to control frame pacing
This is useless
Rename timerUpdate() and timerGetRate() to timerUpdateClock() and timerGetClockRate(), because they both only apply to totalclock
Use nanosleep instead of usleep
Patch from Doom64hunter to address a couple of issues with per-map art
Renderer sprite sorting patch from Nuke.YKT
Explicitly implement timerGetTicksU64() with chrono::high_resolution_timer and use chrono::steady_clock for regular timerGetHiTicks() and totalclock
Recalculate frame limiter delay every time a frame is drawn
This is an attempt to work around issues that arise when timerGetTicksU64() isn't monotonic. Timing situations like that shouldn't happen, but some systems have pretty messed up configurations.
Let's try something different for r_swapcomplete
Remove obsolete win_uninit() function
Disable SDL's built-in signal handler
Call debug_break() from attach_debugger_here() in builds with DEBUGGINGAIDS enabled
This old 3dfx and SiS hardware isn't capable of running our GL renderer and never will be
Equivalent to jfbuild commit b6a666745cea21c6f8193b046bcdd0636fb7c166
Simplify SDL subsystem init a little
Rename BUILD_xxx debug env vars to EDUKE32_xxx
Add EDUKE32_NO_NVIDIA_THREADED_OPTIMIZATIONS for Linux and EDUKE32_NO_XINPUT for Windows
Don't send a mouse event when clicking to re-focus the game window
Fix oversight where the check for whether the loaded version of SDL is new enough was never updated for SDL2
GLU isn't loaded with SDL, so calling SDL_GetError() here is pretty useless
Use aligned allocation for the block gamecontrollerdb.txt is read into
I know, unnecessary.
SDL_free() is not for pointers to SDL_RWops structures...
Don't use the SDL joystick API with SDL2, just the updated one for game controllers
Add optional duration parameter to idle()
Clamp nextPageDelay in G_FPSLimit() to 0.0 - g_frameDelay
Increase timer resolution on Windows
Remove win_allowtaskswitching()
We weren't ever calling win_allowtaskswitching(0) anywhere, and when I tried it, it did not work. Broken functionality to try and block alt + tab is not necessary.
Use timeGetTime() to implement timerGetTicks() on Windows
Remove dead function declarations
This function declaration was in the wrong header
Update xxHash to git commit cd0f5c22095c49c35104dca86620322801f14893
Use timerGetHiTicks() instead of timerGetTicks() for timing in osdfunc_fileinfo()
G_HandleAsync() -> gameHandleEvents()
Partial timer cleanup
After this revision, the only place timerUpdate() is called is from within handleevents(). S_Cleanup(), MUSIC_Update(), and G_HandleSpecialKeys() are now called from a timer callback set with timerSetCallback(). This more or less deprecates the usage of faketimerhandler() in EDuke32 and Mapster32, but other games still rely on the functionality.
Reword stupid error message that has probably never even been seen
ifdef DEBUGGINGAIDS for this quick exit crap
this is more like regular AIDS than DEBUGGINGAIDS though