Build 9879
GIT-
fix building against SDL older than 2.0.9
December 2021 4 builds · 17 commits
Build 9875
GIT-
extra const in libtess2
-
fix use of value before validation in polymer_updatewall()
-
prevent potential deferencing of null ptr
-
this sort of type punning isn't necessary anymore
-
fix copy/paste error in screentextRender()
Show details
For some reason, this error didn't propagate into screentextGetSize().
Build 9839
GIT-
tiny polymost optimization; swap 4 floating point divisions for multiplications in polymost_bindPth()
Build 9831
GIT-
fix more stupid errors in b6740a7b32d352e6d71a5be0bc118f381599933e
November 2021 9 builds · 28 commits
Build 9828
GIT-
fix dumbass GL state oversights
-
fix stupid copy/paste error in b6740a7b32d352e6d71a5be0bc118f381599933e
-
work on controller features
-
fix oversight in current A_RadiusDamage() implementation that allowed a sprite to inflict radius damage on itself
-
fix animvpx
Build 9815
GIT-
allow r_pr_shadowdetail 6 (8192x8192)
-
add missing buildgl_bindSamplerObject() calls
-
add SAMPLER_NEAREST_NEAREST_CLAMP and SAMPLER_NEAREST_NEAREST_WRAP, samplers which use GL_NEAREST_MIPMAP_NEAREST for the minification filter
Show details
It looks like mipmapping actually works with the indexed color texture shader in Polymost as-is, but it doesn't work with Polymer art mapping so we're disabling it there.
Build 9804
GIT-
oops, this still needs to be called here for Polymer to work in the editor
-
fix warning
Build 9791
GIT-
add amalgamated version of the vector math library found at https://github.com/glampert/vectormath
Build 9787
GIT-
clean up r_persistentstreambuffer so you can turn it on and off without having to restart anything
-
move all the cvars in polymost.cpp so that they're at least in the same spot
-
fix a few polymer lighting bugs, add interpolation to light positions, etc
-
improve GL state management, add support for GL sampler objects, add libtess2 and implement in polymer in place of the GLU tesselator, drop dependency on libGLU, improve GL debug output (in debug builds), fix various GL resource leaks
Show details
This should prevent a few more state changes, and address issues arising from rendermode changes wiping our state but not any of our accounting data for said state. Because libtess2 seems to be abandoned and because there were a few issues I needed to fix anyway, I went ahead and made a custom single-file header-only version for improved performance.
Build 9782
GIT-
detect GL_ARB_sampler_objects, drop detection of GL_ARB_vertex_buffer_object, GL_EXT_texture_edge_clamp, and GL_SGIS_texture_edge_clamp
-
remove unfinished x86_64 port of the original x86 classic renderer
Show details
It never worked.
Build 9767
GIT-
move yax_updown[] reset to yax_update()
-
don't use SDL_WINDOW_FULLSCREEN_DESKTOP and don't call SDL_SetWindowBordered in fullscreen
-
ignore SDL_WINDOWEVENT_MOVED and SDL_WINDOWEVENT_RESIZED in fullscreen
Show details
I'm not sure if this is going to end up being the correct thing to do here, but it's what we shipped in the Fury 2.0 binary.
-
better multi-monitor support
-
fix console logging issue on Linux introduced by logging rework
Show details
TIL vsnprintf and _vsnprintf are not, in fact, the same thing.
Build 9751
GIT-
add "notrans" .def token for voxels to disable drawing with translucency in GL modes
Build 9748
GIT-
address defect in c653ba7fc09257ed3590cd2ce29d155ef6752032
October 2021 6 builds · 49 commits
Build 9747
GIT-
add libasync (async++), https://github.com/Amanieu/asyncplusplus/
-
update libdivide to b322221677351ebb11f0a42fe9a9a2794da5bfe5, add libdivide_config.h
Show details
engine: update libdivide to b322221677351ebb11f0a42fe9a9a2794da5bfe5, add libdivide_config.h
-
print commit hash from SDL_GetRevision() to log at startup when using SDL dev builds (odd-numbered patch level)
Show details
This is admittedly terrible, but it shouldn't break unless the SDL_GetRevision() output format changes, in which case it just won't print the string anymore.
-
prune SDL_mixer stuff from sdl_inc.h
Show details
We will never NEED_SDL_MIXER.
-
fix tabs in random.h
Build 9721
GIT-
General
Revert "engine: make floor sprites clip a little more like floors"
Show details
clipmove() really needs concrete indication of a moving point's starting floor height for this optimization to work correctly, and inferring that ourselves without it being explicitly passed in isn't something that would work.
- Reverts
- 81f7c71 — engine: make floor sprites clip a little more like floors (Build 9710)
Build 9720
GIT-
fix SDL fullscreen issues
-
remove r_pr_vbos cvar and the code paths used for values 0 and 1
Show details
r_pr_vbos 0 was broken and r_pr_vbos 1 became useless once it was clear r_pr_vbos 2 wasn't causing issues.
-
add GL context version to glinfo, allow in 8 bpp mode (it's still backed by a GL surface, after all...)
-
manually flush console log file ptr in crash handler
Build 9710
GIT-
add coordinates to log output from clipmove()
-
make floor sprites clip a little more like floors
-
bump MAXCLIPNUM to 4096
Show details
This isn't really ideal, but neither is hitting the limit due to maps with an excessive number of sprites that use clipshapes.
-
fix double free on exit
-
implement buildvfs_exists() using GetFileAttributes() on Windows and stat() elsewhere, FORCE_INLINE some stuff in vfs.h
Show details
The rationale behind this change is that I was profiling CPU use during loads of particularly heavy .def files (10000+ items) and discovered an inappropriate amount of CPU time was being spent inside access(). So, I did some research and found several cases where people on stackoverflow etc had asked what the fastest method of determining whether or not a file exists was, with the results of several benchmarks being given as answers. GetFileAttributes() seemed to win on Windows and stat() seemed to win on Linux, and I did not see any results where access() was a serious contender when included in the spread of tested functions, with the best I saw being 1100-something nanoseconds for access() and 800-something for stat(). This whole thing would be faster if we just read in the entire directory index and matched files against it ourselves with just the few API calls involved with retrieving the index, but access to uncooked data littered across a directory tree isn't really something I give a shit about optimizing further at the moment.
Build 9679
GIT-
fix incorrect lower bound for r_anisotropy
-
inline openfrompath_internal
-
disable SDL_HINT_WINDOWS_DISABLE_THREAD_NAMING if _MSC_VER is defined
-
include mimalloc-new-delete.h in smmalloc_generic.cpp
Show details
This defines global new/delete overrides, so look here if problems stemming from those ever arise.
-
don't crash when loading corrupt maps with sectors with zero walls
Show details
This also makes the editor recognize corrupt sectors with < 3 walls and try to repair the map. It works for the corrupt sectors in E4L1, at least. Please immediately report any problems caused by this change directly to me.
Build 9631
GIT-
General
engine/mact: reimplement controller dead zone so that it isn't just naively applied to each axis independently
-
rename r_windowpositioning var in source, default to 1
August 2021 2 builds · 27 commits
Build 9584
GIT-
allow pasting into the console from the system clipboard with RMB
Show details
Disable with "osdclipboard 0".
-
clear mouse buttons when toggling the console