Dicembre 2019
-
Generale
Recalculate horizycent because ydim can be changed in renderSetTarget
-
Generale
Revert "Windows: Remove keyboard layout switching entirely"
Mostra dettagli
This change was not well thought out and breaks a number of things for users of foreign keyboards. Unfortunately, we are still at a point where removing keyboard layout switching is a QOL downgrade.
-
Remove keyboard layout switching entirely
Mostra dettagli
Upon testing this already seems to do the right thing in game and editor, with winlayer, SDL 1.2, and SDL 2: use the labels on the keys when entering text such as in the console, and the positions of the keys for everything else.
-
Generale
Remove -msse2 from default 32-bit x86 optimization parameters
-
Generale
Improve diagnostic printing of slopalookup overflows
-
Generale
Fix regression in software rendering of slopes in r8363
-
Generale
Fix SDL 1.2 build
-
Generale
Fix RENDERTYPE=WIN build
Mostra dettagli
Broken in r8319.
-
Generale
Amend r8174: Remove call to windowsHandleFocusChange during SDL_WINDOWEVENT_ENTER and SDL_WINDOWEVENT_LEAVE
Mostra dettagli
These events are for when the mouse cursor enters and leaves the window's region, not any kind of focus change. Confirmed with a test app. The SDL Wiki's description, "window has gained/lost mouse focus" is misleading.
-
Generale
Fix crash when specifying an out of range value for sys_timer in the console
Mostra dettagli
Manifests as a crash on startup for all users who previously had sys_timer set to 4.
-
Generale
This should fix rare stalls of several seconds when changing window focus on some Windows systems, related to changing the keyboard layout
-
Generale
Fix SW compilation under VS
-
Generale
Fix Mapster32 undo corruption issue
-
Generale
This is a __restricted area
Mostra dettagli
Theoretically, this will help the compiler generate more optimized code. In practicality, I haven't seen any difference yet but I don't think giving the compiler hints as to the intended operation of the code is ever really a bad thing.
-
Generale
Track cache1d block overhead differently
-
Generale
These shouldn't be references
-
Generale
Update cvar help strings, WIP
-
Generale
Improve console cvar printing
Mostra dettagli
This breaks cvar printing out of osdcmd_cvar_set() and creates osdfunc_printvar(), which prints the cvar's current value and the range of values it accepts. This also improves the output of the "help" command.
-
Generale
Remove dead code
-
Generale
Fix cacheAllSounds() being called before .def parsing... oops!
-
Generale
Cast the nanoseconds to uint64_t before multiplying in timerGetTicks()
-
Generale
Move logic for timer type selection into its own function
-
Generale
Use -march=nocona for x86_64 builds
Mostra dettagli
This appears to be preferable for producing generic x86_64 binaries these days.
-
Generale
Use -march=pentium-m for i686 builds
-
Generale
No more std::chrono!
-
Generale
Timer maintenance
-
Generale
Fix osdcmd_changelevel
-
Generale
This change to Gv_DivVar() is faster when benchmarked
-
Generale
Clean up CON_FOR a little
-
Generale
Change VS project DiagnosticsFormat option
-
Generale
Only use lfence with rdtsc
Mostra dettagli
I was going to use the CPU vendor to determine whether to use mfence or lfence, but reading the links in the comments lead me to the conclusion that lfence is still the correct choice on AMD.
-
Generale
Start ripping out std::chrono because it sucks
-
Generale
Remove sound init messages from Shadow Warrior as they're emitted by the library itself now
-
Generale
Add snd_sdl_audiodriver cvar to select which audio backend SDL uses
Mostra dettagli
This also cleans up the sound init messages a bit.
-
Generale
I'm pretty sure none of our output goes through the code path that uses this hint, but if it does we probably want the good one and not the crappy one
-
Generale
Don't write blank string cvars to settings.cfg
-
Generale
Remove driver_nosound from audiolib
Mostra dettagli
Why? Because it's fucking useless. If we want to waste CPU on outputting nothing, SDL has us covered with its "dummy" audio backend.
-
Generale
Remove obsolete option to reverse the stereo channels
Mostra dettagli
Anyone who plugged in their speakers backwards is SOL.
-
Generale
ENet fix to allow running on Windows XP again... completely untested tbh
Mostra dettagli
This is not the only thing that breaks compatibility with XP, but this fixes the ENet part of it.
-
Generale
Add mus_winmm_device cvar to allow selecting the Windows MME MIDI output device
-
Generale
Add mus_xmp_interpolation cvar to control XMP interpolation mode
Mostra dettagli
Full disclosure: changing the interpolation mode is practically useless. I was going to drop this commit from my tree, but a later commit doesn't merge without it and I didn't feel like dealing with it.
-
Generale
Only restart the sound system itself when pressing "apply" in the advanced sound setup menu if the playback rate or number of voices changed
-
Generale
MIDI shuffling
-
Generale
Use mouse wheel to scroll console and limit scroll area to the number of lines actually present
-
Generale
Use GrowArray<> to save individual lines of text printed to the console so we can rebuild the concatenated text buffer without fucked formatting when the console resolution changes
Mostra dettagli
A proper solution would probably shitcan the big buffer altogether and just print the lines stored in the GrowArray<> directly, but this commit is not that.
-
Generale
Update console cvar descriptions
-
Generale
Change osdlogcutoff to control the maximum number of errors printed to the console
Mostra dettagli
Previously osdlogcutoff controlled the max number of lines (error or otherwise) printed to the console, and the error limit was hard-coded at 4096. This seems more useful.
-
Generale
intitialization
-
Generale
Remove duplicate call to cacheAllSounds() from S_SoundStartup()
-
Generale
Move atoi_safe() from editor.h to compat.h and #define Batoi atoi_safe
Mostra dettagli
They were the same thing so they might as well be in the same place.