black [Wed, 2 Feb 2005 19:30:16 +0000 (19:30 +0000)]
Added support for color tags to the console. Use the @ char followed by a
number to switch the color. Currently I have only the color indices 0-7 are
valid.
havoc [Tue, 25 Jan 2005 20:05:18 +0000 (20:05 +0000)]
now supports Mac and Windows newlines everywhere
all text files are now saved in UNIX format
removed FS_Gets and FS_Getline (they were only used by the loadgame code along with sscanf... bad)
added FS_UnGetc (needed for Windows newline support in FRIK_FILE fgets)
havoc [Mon, 24 Jan 2005 21:51:16 +0000 (21:51 +0000)]
added hacky workaround for nvidia GLSL bug (state screwups after switching back to program 0 which go away if I draw something (even an empty Begin...End) after the switch)
havoc [Sun, 23 Jan 2005 22:16:33 +0000 (22:16 +0000)]
GLSL shader per pixel lighting implemented (with optimized permutations)
offset mapping implemented in GLSL lighting shader (looks really cool if a little slow, enabled by default)
cleaned up some internal handling ambientscale/diffusescale/specularscale/gloss mode design issues in R_Shadow_RenderLighting
havoc [Sun, 23 Jan 2005 22:12:22 +0000 (22:12 +0000)]
rearchitected renderer backend to have support for independent number of texture image units, texture coord array units, and texture env blend units
added GL_Backend_CompileProgram and GL_Backend_FreeProgram for GLSL support
havoc [Sun, 23 Jan 2005 06:39:07 +0000 (06:39 +0000)]
fixed blendfunc issues with single pass lighting (such as ambient pass) which were using GL_DST_ALPHA when they shouldn't have been (fixes the 'bright shadows' issue which occurred from multiple lights in some cases)
added a usestencil variable in rtlight rendering just to make it a little more clear in the code
havoc [Fri, 21 Jan 2005 03:22:11 +0000 (03:22 +0000)]
changed GAME_ init data to have a separate user config directory for each game type (~/.darkplaces for most), and two game dirs rather than an implied id1 (so now nexuiz doesn't look for id1 for example)
molivier [Wed, 19 Jan 2005 09:09:20 +0000 (09:09 +0000)]
- Got rid of all the FILE* stuff in the FS code, relying on low-level
functions exclusively. It now manages its own per-file 2KB read buffers.
Writes are no longer buffered.
- Removed function "FS_Flush" and cvar "log_sync", since they're no
longer relevant
- Package's file descriptors are now duplicated when opening a new file
inside a package, to make package substitution at run time impossible on
non-Win32 systems. Added a couple of "lseek" when necessary (before
"read" calls in "FS_Read").
- Removed unused functions "FS_Eof", "Log_Print" and "S_ClearBuffer"
- Several other minor changes and fixes in the FS code
- Changed log open mode to "a" from "at" since 't' seems a Win32-only
extension
havoc [Mon, 17 Jan 2005 23:30:20 +0000 (23:30 +0000)]
now adapts sv.num_edicts according to how many edicts are found by SV_Physics, so if a lot of temporary stuff spawns and removes itself it will knock the num_edicts back down
havoc [Sat, 15 Jan 2005 02:11:53 +0000 (02:11 +0000)]
fixed a bug in texture generator that made all the particles too large in the particle font by one pixel (now they're smaller than their frame, like they were supposed to be), this fixes all the sharp edges on particles
changed nexbeam to be quite a bit brighter in the center than the sides (as I really wanted it all along)
nexbeam no longer scrolls its texture (looked odd)
havoc [Thu, 13 Jan 2005 06:36:57 +0000 (06:36 +0000)]
shuffled around some SCR_BeginLoadingPlaque calls so now SV_SpawnServer causes a loading plaque and ParseServerinfo only draws a loading plaque if there's no local server (which already did one), this cleans up the loading plaque management a great deal
havoc [Thu, 13 Jan 2005 06:29:28 +0000 (06:29 +0000)]
NetConn_Read no longer runs cl_packetloss rejection code if there is no packet (slight optimization/sanity improvement)
removed another SCR_BeginLoadingPlaque call
havoc [Thu, 13 Jan 2005 06:11:38 +0000 (06:11 +0000)]
bound the fractions in CL_TraceLine instead of posting a warning, as out of bounds fractions have no harmful effects during a trace (to allow the bounds in the trace code to be removed)
havoc [Thu, 13 Jan 2005 06:07:23 +0000 (06:07 +0000)]
rearranged entire SV_Physics process (which may break mods but I hope not!):
each operation (move, prethink, think, postthink) is now a separate stage (as an optimization mainly)
moves are now done before prethinks, thinks, and postthinks so that think-spawned entities do not move on their first frame (this fixes the issue that all projectiles appeared some distance from the player, dependent on server framerate, so rocket trails and other things look correct)
added a relink of any entity that moves for any reason (just an origin compare), not sure if this is a good idea but it is more consistent
havoc [Thu, 13 Jan 2005 05:55:46 +0000 (05:55 +0000)]
don't generate vertex/normal/tangents until they are needed by lighting passes, this greatly improved framerates in nexuiz because a lot of models were having nothing to draw (apparently), perhaps their light color was black (odd...)
molivier [Sun, 9 Jan 2005 09:07:08 +0000 (09:07 +0000)]
- Sys_LoadLibrary now accepts several possible names for a DLL (it fixes
the PK3 and Vorbis supports on NetBSD)
- Added a check for absolute paths in FS_CheckNastyPath
- Moved "COM_Shutdown" after "Log_Close" in the shutdown process to make
sure the FS code is still active when we close the log file
- Factorized some code in the FS module
- Made the "listdirectory" parameter "const"
- Fixed a warning in "prvm_cmds.c" when compiling with GCC
havoc [Thu, 6 Jan 2005 13:31:31 +0000 (13:31 +0000)]
swapped order of incr/decr renders for static shadow volumes (for some reason I had decr first, then incr, doesn't matter since they don't clamp but looks odd in the code)
remove quotes on cubemap names in .rtlights file loader (forgot to commit this previously)
black [Sat, 1 Jan 2005 19:00:33 +0000 (19:00 +0000)]
-Added 2 builtins to make the hostcache stuff easier extensible.
More to follow soon.
-Cmd now has a shutdown function so it doesnt leak anything (hopefully).
black [Tue, 28 Dec 2004 22:29:45 +0000 (22:29 +0000)]
-Changed the name of PingTest to Test (it now also adds a name to the entry)
-Now the test function is enabled by default (just so Vermeulen can play around
with it) - there shouldnt be any conflicts with real servers since I shift the
ping of all emulated servers by 250 ms.
black [Mon, 27 Dec 2004 16:54:32 +0000 (16:54 +0000)]
-Added support for 10 hostcache masks, which will be at the same time.
5 will be concatenated with AND and the other 5 with OR, so you'll have
plenty of possibilities to mask whatever you want.
-Changed the mask structure and function to be a bit more generic.
-Added 3 new comparison operators (NOTEQUAL, CONTAINS and NOTCONTAIN - the
latter two for strings)
-Adapted the builtins to work with multiple masks.
havoc [Thu, 23 Dec 2004 00:18:03 +0000 (00:18 +0000)]
changed r_bloom_resolution behavior to have a non-square texture to match the view width/height aspect
changed default r_bloom_resolution from 256 to 320
changed r_bloom_blur to adapt to resolutions (using 320 as a base value - blur 8 is 8 pixel blur at resolution 320, 16 pixel blur at 640, etc)
black [Wed, 22 Dec 2004 23:10:20 +0000 (23:10 +0000)]
-Fixed a wrong parameter type in netconn.
-Made hostcache_cachecount public again.
-Added altstr_set to the new VM.
-Started work on the new hostcache interface for the menu qc.
black [Tue, 21 Dec 2004 21:41:58 +0000 (21:41 +0000)]
-Added a new hostcache interface:
* Ascending und descending sorting by various fields is supported.
* Its possible to mask entries (substrings or comparisons for numbers).
molivier [Mon, 20 Dec 2004 09:58:41 +0000 (09:58 +0000)]
- Fixed JPEG support on Mac OS X
- Fixed Ogg Vorbis support on Mac OS X (the vorbis DLL is now also loaded manually since some OSes don't do that for us)
- Fixed loading of MD3 models on big-endian CPUs
black [Sun, 19 Dec 2004 22:55:00 +0000 (22:55 +0000)]
-Fixed a "bug" in PRVM_iscachedpic - it always returned true and thus no
images were precached (now it always returns false and the menu halts for
~4 seconds after starting)