2008-04-04 Various contributors + Solaris fixes + Replace vsprintf function in bg_lib.c with vsnprintf implementation started by Patrick Powell + Compile bg_* files separately for each game module, as originally intended + Write q3config_server.cfg for the server, to avoid reseting client variables after running a dedicated server + Split image decoders into their own files + OS X build updates for most compatibility + Slackbuild + Detect available resolutions and offer them in the in game menus + A few botlib fixes + Fix poppy captured audio when recording videos + Extend console logging on crash errors + Merge *BSD platform definitions in q_platform.h + IRIX support + Remove all the old bat/sh QVM building scripts + Make master server used client configurable (cl_master) + Fix to QVM compilation on big endian architectures + OpenBSD support + Autocomplete key names + Don't build client command completion on the dedicated server + Don't apply colour escape chars on input fields + Rewrite of the win32 dedicated console + Improved Makefile startup time + Build dedicated server binary on Windows + Bump Q3_VERSION to 1.35 + Replacement of platform specific backends with a generic SDL one + Merge win_net.c and unix_net.c to net_ip.c + Demote input related console information to developer only so that it doesn't spam the console every time input settings are changed + PNG texture support + Cleanup of tabulation in R_LoadImage + Fixes to console scrolling + New x86_64 vm that doesn't use gas + Early out AABB collision optimisation + Generate QVM dependicies in a better way + Build process is quieter + Replace horrendously long list of Makefile build rules with set of inference rules + Allow CC to be overridden externally to the Makefile + Move storage of console history from a cvar to a file in order to alleviate security concerns + Fix bug where transparent surfaces wouldn't draw over skyboxes + Add input sanitising to various sound playing functions called from mods + Explicitly set OpenAL distance attenuation model + Increased the number of registers used for the opStack in the PPC vm from 12 to 16 + Fix endian issue in MDR loading + Add cURL support for HTTP/FTP downloading + Disable video command when not playing back a demo + Print the SVN version string in Com_Init() + OpenAL device enumeration support + Fix 100% CPU usage on idle dedicated servers + Windows home directory support + Improve correctness of AVI files created by video command + Better SDL joystick support + sv_minRate + [cl|sv]_packetdelay + Various security fixes + Fix JIT compiler code execution on NX-protected win32 OS + Fix r_overBrightBits variable getting ignored on Linux + cl_guid for semi-reliable server authentication + Anisotropic texture filtering + Video export doesn't crap out with sv_pure 1 anymore + Video export doesn't crap out when writing > 2Gb files anymore + Fix to a bug where servers with long uptimes (~27 days) would consume 100% CPU if the running game did not set the nextmap cvar + Some OSes no longer requires a vid_restart when changing r_fullscreen 2006-01-24 Various contributors + Persistent console history + Added code to sleep a bit when q3 has no focus and sleep a lot when it's minimised (SDL only) + Cull excess speaker entities when using OpenAL + Fix the operation of the delete key in *nix + Only check the checksum on baseq3 pak0.pk3 + Overhaul of console autocompletion - No longer does weird stuff like move the cursor inappropriately - Autocomplete works with compound commands - Special autocomplete on some commands e.g. \map, \demo - Removed various hacks used to counter the original autocomplete code + Fixed the ability to disable Ogg Vorbis + s/i686/i586/ - see bug #2578 + Some sloppily coded mods call the Q3 sound API with NaNs -- sanitise this + Removed advertising clause from BSD license as per mailing list discussion + "make distclean" now does what you'd expect + "make clean toolsclean" now does what "make distclean" did before + GPL MD4 implementation 2006-01-16 Various contributors + Move code/unix/Makefile to ./Makefile + x86 OS X support + "quake3" shell script as shipped with 1.32 (on linux) no longer needed + Ogg codec support from Joerg Dietrich + Fix to the gcc4/-O0 x86 JIT compiler bug + Up the defaults for zone and hunk memory since some mods (UT) have large memory requirements that will have increased versus 1.32b due to some of the alignment fixes + Dependency generation for the .asm files + Remove FS_SetRestrictions + Add FS_CheckPak0 for better error messages where dumb users are involved + Added cl_autoRecordDemo, which when enabled automatically records a new demo on each map change + Only display the g_synchronousClients warning when it's appropriate + Remove custom memcpy/memset code + AVI video output - Uses motion jpeg codec by default - Use cl_aviFrameRate to set a framerate - \video [filename] to start capture - \stopvideo to stop capture - Audio capture is a bit ropey + General Makefile improvements + Support for MinGW cross compilation + NetBSD support from optical + x86_64 JIT bytecode compiler no longer disabled by default + msvc project files updated and moved to win32/msvc + Various alignment fixes + Solaris (x86 and sparc) support from Vincent S. Cojot + Fixed Altivec-based mesh rendering + Ditch Mac OS 9 support + Added a Makefile option USE_LOCAL_HEADERS which can be disabled to use system headers if desired + Detection of Altivec on Mac OS X + SMP support with sdl_glimp.c on Mac OS X. + Add "very high quality" option (patch from Pascal de Bruijn) + Support for RIFF files with zero length data chunks (yes they exist, and yes, they're legal) + Support for ccache. If you want it, add USE_CCACHE=1 to Makefile.local + Mac OS X now uses SDL backend, all Objective C removed + Partial implementation of FS_Seek for files in pk3s + Implementation of r_dlightBacks from Shane Isley + OpenAL support, from BlackAura aka Stuart Dalton + An abstract codec system, simplifying support for new formats + Ignore in_dgamouse setting if dga isn't available + Removed hard coded mouse acceleration in *nix input code + Basically rewrote the lcc Makefile to be more sane + Removed various bits of lcc that weren't built/needed + General portability improvements + Various variables added that aid packaging, from vapier + Centralise architecture defines in q_platform.h + Replaced a bunch of inline and __inline with ID_INLINE + Replaced a bunch of __i386__ with id386 + General tidy up of asm preprocessor decisions + Removed C_ONLY from the dedicated server build + Removed rule to build C++ (for splines) from the Makefile + General decrufting + Split USE_SDL into USE_SDL_VIDEO and USE_SDL_AUDIO + Various assorted bug fixes 2005-10-29 Various contributors + nasm syntax asm ported to gas + Disabled-by-default MD4 support + cons build system removed + Better FreeBSD support + Makefile generates dependencies + Some SDL sound tweaks + qvm build tools and qvms are now built with the rest of the binaries + q3asm-turbo from Phaethon + Moved various displaced c and h files into more appropriate places + A shitload (can I say shit?) of bug fixes -- see the svn log for details 2005-09-22 Tim Angus + MinGW port 2005-09-?? + SDL Stuff (icculus) + x86_64 (ludwig von angstenheimer) + patches from a cast of thousands 2004-05-22 Timothee Besset + updated the xcode project from Apple's version now with the latest vm_ppc code 2004-05-21 Timothee Besset + fixed the Linux build to compile again on sid (glext.h and gcc3 warnings) + 2 weeks ago, hacked up the source to compile on panther / xcode 1.1 several cleanups were needed, and VM support seems broke (hangs or crashes) + got altivec optimisations from Apple (Kenneth Dyke) merged back in + looks like with the new code merge the VM support is back in and working 2003-09-15 Timothee Besset + import Q3 java master code, cleanups on monster 2003-08-31 Timothee Besset + loki_setup hell https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=626 http://zerowing.idsoftware.com/linux/q3a/index.html#glibc text mode installer in loki_setup image built on Mandrake 7.2 crashes on some glibc 2.3 systems such as RH9 etc. need to move to a different version of the installer, and update old installers to keep them still 'installing' moving to build the setup binaries on Debian Woody systems (glibc 2.2, text mode installer will no longer work on 2.1 systems) hacked together a new setup, using setup tree from RTCW. would need a complete revamp if a new full setup with new binaries is needed 2003-07-17 Timothee Besset + new cvsreport, testing per-module config 2003-01-19 Timothee Besset + building on both gcc 2.x and 3.x added conf modules to check gcc version ccache support 2003-01-13 Timothee Besset + tweaking around for gcc 3.x build edit Conscript to change the compiler 2002-12-16 Timothee Besset + added pbEmit class to auth code, emit CD keys to local PB master 2002-11-14 Timothee Besset + up to latest makeself.sh + add both quake3.x86 and quake3-smp.x86 to setup + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=573 console setup crash / glibc 2.3 (Debian Sid) investigated, put together a workaround 2002-11-5 Timothee Besset + Linux building both smp and non-smp again. Will have to put both in setup + added in_subframe to toggle X subframe event handling + reworked the timing code to be more reliable + cleaned up dgamouse/in_mouse code, removed unnecessary dgamouse var + made the mouse grabbing an in_nograb cvar, no longer a compile time option in_nograb 1 force in_dgamouse 0 and r_fullscreen 0 (any of those two will b0rk) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=565 mouse issues on Suze 8.1 - related to subframe event timing added code chunk to detect broken X timing and disable subframe + tweaked the subframe/X bug workaround to be less paranoid 2002-10-28 Timothee Besset + no longer blocking demo recording if g_synchronous clients != 0 only sending out a warning (everyone does g_sync 1 ; record ; g_sync 0) 2002-10-21 Timothee Besset + building final mod sdk setups (added lcc bins, added link to q3asm-turbo in readme) 2002-10-8 Timothee Besset + quickfix cl_maxpackets > 125 brings back to 100 1.32 release --- 2002-10-7 Timothee Besset + made the 'demo' command case-insensitive on extension match (it was confused by demo FOUR.DM_68) + mouse wheel scrolling with in_mouse 1 + window mode was not working, fixed (DI didn't catch) + removing on-the-fly pk3 build from Linux setup, using the finalized ones now added 'pk3' option to cons for toggle of pk3 building 2002-10-5 Timothee Besset + updated win32 mod sdk (in win32/mod-sdk-setup) added q3asm and lcc source updated the .bat to build VMs 2002-10-3 Timothee Besset + linux mod sdk, wrote the bulk of the scripts 2002-9-30 Timothee Besset + ATVI Quake 3 1.32 Patch #9 rolling back to the way it was before, leaving 1v1 force vote exploit, the fix was worse than the bug from comment on bug #9 in tracker: actually the fix is worse than the original bug after the fix, voting when you are alone on the server was no longer working it was kinda intended in the fix, that you would have to be at least two to pass a vote .. but it is an oversight. calling a vote in a 1v1 game against a bot fails immediately (calling a vote in any situation where there's only 1 live player fails) Say a server's running some lame custom map that you have but a friend doesn't. You can't go on the server and change it to the map you want to play, so he ends up having to auto-dl it at 8K a second just so you can switch from it. This particular 2 clients, vote / quit exploit would involve too many changes to fix properly. I am reverting back to the old version, and leaving as WNF 2002-9-29 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=50 added Wheel support to the DirectInput code IN_DIMouse (in_mouse 1) tweaked the Wheel mouse reading for in_mouse -1 (old win32 input code) handle correctly when zDelta is > 120 provide a in_logitechbug cvar to handle buggy Logitech MouseWare driver sending wheel events twice 2002-9-26 Timothee Besset + ATVI Quake 3 1.32 Patch #38 adding trap_SetPbClStatus, reliably checks for PB presence before enabling PB in UI 2002-9-25 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=551 SVF_CLIENTMASK, fixed a typo + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=555 pushed cl_maxpackets upper limit to 125 (from 100) per CPMA Arqon's request 2002-9-24 Timothee Besset + ATVI Quake 3 1.32 Patch #33 PB reporting sv_paused cvar hacked, fixed SV_CheckPaused to use a Cvar_Set + ATVI Quake 3 1.32 Patch #24 added [skipnotify] from RTCW, use to display in the console only, but not on client screen (also fixes pb_msgprefix and pb_sv_msgprefix) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=553 using correct error message if listen server starting as cl_punkbuster 0 sv_punkbuster 1 + ATVI Quake 3 1.32 Patch #35 text auto wrap in UI code was eating the last word if it was wrapping fixed in Q3 and TA UI (this bug could have affected the server print message also) + some updates to the win32 cons post-build process 2002-9-21 Timothee Besset + adding bspc cons build script 2002-9-19 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=552 disconnect reason is transmitted in the disconnect command and processed into com_errorMessage (similar to RTCW behaviour) added UI for com_errorMessage cvar in baseq3/, if client is kicked/dropped/disconnected for whatever reason (this is already functional in TA) + ATVI Quake 3 1.32 Patch #9 failing vote if there's only one voting client (fixes exploit on 2-player server where one player votes then disconnects, forcing vote to pass) + ATVI Quake 3 1.32 Patch #5 removed the userInfoChanged message (was a debugging leftover) + ATVI Quake 3 1.32 Patch #18 rcon was not properly fixed yet, this only showed up for PB commands changed the rcon parsing again to be more reliable 2002-9-18 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=549 the demo command has a list of compatible protocols, it will loop through 66 67 68 you can do '/demo four' and it will try four.dm_66 four.dm_67 four.dm_68 or you can explicitely give a '/demo demoname.dm_??' + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=551 added SVF_CLIENTMASK (0x00000002), works only with <= 32 players set bitmask of players to which send entity 2002-9-17 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=550 rcon bug fix + some scons updates for win32 (post build) + 1.32rc2 2002-9-06 Timothee Besset + updated completely the setup system: fixed cons stuff to build setup with cons -- release setup working from new setup codebase with some custom patches: https://bugzilla.icculus.org/show_bug.cgi?id=52 https://bugzilla.icculus.org/show_bug.cgi?id=53 checked that BSD support was still in (brandelfing and symlinks) .. will have to get tester feedback bumped version to 1.32rc1 TODO: update the windows .VCT (standalone setup and auto-update) 2002-9-04 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=543 backport from RTCW 1.4 code rcon commands where sent after being tokenized and rebuilt that was breaking any quoting, for instance 'rcon g_motd "hooka pooka"' added Cmd_Cmd() to retrieve the un-tokenized command and transmit as is on both ends + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=542 b0rked text wrapping in connect screen was a missing sizeScale in q3_ui/, and a bad param in ui/ + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=540 backport fix to pk3 reordering, happens when clearing the references, bad order from connection may break stuff + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=527 TA ui/, quickfix to netSource (mod stuff, doesn't affect TA) + cleaned up broken old DO_WIN32 stuff in cons scripts + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=526 typo in models2.shader + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=443 Linux client: sub-frame timing of key/mouse events + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=453 added mousewheel support: wheel to scroll, ctrl+wheel to scroll faster, shift+wheel to scroll history + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=545 bumped server count to 4096 + keep around: __asm__ __volatile__ ("int $0x03"); + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=516 moved screenshots to backend with a new RC_SCREENSHOT render command fixes the r_smp 1 garbled screenshots 2002-8-29 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=539 new VM code from Raven's Sof2 + cons / qvmtools build system fixes + had to get a new qe3.ico again (resource compiler error) http://vasin.hypermart.net/eei.htm + updated, basic testing on win32, merging back in trunk + merged bug-539 branch back into trunk, officialize the new VM code 2002-8-26 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=472 linux client: handle ctrl+space situations (could leave space locked on + space not working with ctrl on) + update the build system, build q3lcc and q3asm etc. on demand + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=62 fixed invisible players/entities 2002-8-23 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=536 fixing donedl being ignored after autodl if map_restart'ed (propagate from RTCW) ignoring multiple map_restart (propagated from RTCW) + reworked the server 'client text ignored' message to only trigger when there's actually a message that doesn't get to the game VM 2002-8-18 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=528 ydnar: reorg bits in the drawsurf sort index, push MAX_SHADERS to 2^12 + commented out some debug stuff in java auth server + added FAQ item with Linux & BSD patch to handle broadcast on multiple interfaces 2002-8-15 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=534 fixing rcon being broken on NT/XP with > 23 days uptime (or so) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=525 changed the rcon buffer size to avoid overflows and dropping part of the message 2002-8-14 Timothee Besset + hacked in some experimental win32 stuff to the cons files (win32 recognition and pk3 installs .. very very experimental but I needed it for win32 dev) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=521 ui/ and q3_ui/ : added text auto wrapping in the connection screen drawing (server message) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=531 removed the MPlayer stuff from the server browsers + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=505 enabled back the ignore if protocol is != (fixes Wolf servers showing in browser) 2002-8-10 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=500 propagated IP banning fix from RTCW 2002-8-08 Timothee Besset + propagate additional sv_lanForceRate fix from RTCW 2002-8-07 Timothee Besset + added trap_FS_Seek 2002-8-05 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=50 fixed the DI mouse init procedure 2002-8-05 Timothee Besset + removed sv_allowanonymous, was dummy and polluting the serverinfo (sv_allowanonymous was designed to flag wether server was public or not, but that's replaced by g_needpass) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=514 sv_strictAuth (default 1): server variable to control wether strict CDKEY auth should be performed this is required if you want reliable cl_guid on the server extended the getIpAuthorize (server->auth message) syntax sending the fs_game at all times (default 'baseq3'), dummy sv_allowAnonymous 0, strict auth flag NOTE: 1.31 server on baseq3 sends a getIpAuthorize packet like: processing packet: getIpAuthorize -1230824753 217.128.77.195 0 the auth server will mistakenly read fs_game as '0' + TAGGED the master / auth source as pre-1_32 will need to go back to this to comment out all my debugging crap 2002-8-04 Timothee Besset + cleaned master server stuff, client was prompting master.quake3arena.com, server was sending heartbeats to master3.idsoftware.com both point to 192.246.40.56, unified to master.quake3arena.com the MPlayer master, master.quake3world.com doesn't exist anymore, switched it to master.quake3arena.com 2002-8-02 Timothee Besset + added auth server source, reorganized + auth server name / master key optionally set on command line for master and auth servers + auth and master config in build system + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=524 changed default GL driver from libGL.so to libGL.so.1 see LSB 1.2 spec: http://www.linuxbase.org/spec/refspecs/LSB_1.2.0/gLSB/libgl.html + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=480 applying the 'no cp command' experimental fix for beta phase + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=462 backported from RTCW, fix to packet fragmenting emission FIXME: there is some verbose code that we have to take out in the final version (grep for #462) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=475 backported from RTCW, don't get dropped if the server changes map while connecting (ignore outdated cp) + PROTOCOL BUMPED TO 68 2002-8-01 Timothee Besset + Linux: dedicated build was not setting up signal handler like the full client does + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=522 SplashDamage bugfix, now clearing client gentity before GAME_INIT call (instead of after) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=498 fixed NET_AdrToString to print the port as unsigned int (for ports > 1^^15, was showing negative) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=501 maintain IP in userinfo sent to game + checking in master server source 2002-7-31 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=513 https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=506 porting fix from RTCW codebase. client re-orders it's pk3s to scan in the same order than the server this eliminates several 'Invalid .PK3 file referenced' situations (caused by client not referencing the same thing as server) + fixed border remnants in ta ui + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=517 ERR_DROP if PB client off / server on conflict when starting local server + quickfix to q3 ui / punkbuster detect in server browser + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=458 code fix, no more taunt spam + cons install of PB .so files + correct MOD_KAMIKAZE and MOD_JUICED in TA games.log 2002-7-29 Timothee Besset + q3 ui: completed confirmation prompts and messages (added UI_ConfirmMenu_Style & UI_Message) + ta ui: backported 'conditionalopen' from RTCW (conditionalopen ) + ta ui: confirmation prompt for punkbuster enable/disable etc. + added the win32 DLLs to pb/win32/ 2002-7-28 Timothee Besset + ta ui: sv_punkbuster in StartServer menu + ta ui: added cl_punkbuster in server browser + view filters are in a modal dialog + new files: filter.menu menus.txt (pak3.pk3 updated) + fix broken link in Linux FAQ 2002-7-27 Timothee Besset + ta ui: PB display in the browser, in it's additional tab, with sorting 2002-7-26 Timothee Besset + PB UI: for baseq3/ AND missionpack/ q3_ui: Punkbuster: Enable/Disable in server broswer (cl_punkbuster) q3_ui: PB logo, PB Yes/No in browser (TODO: validate this to be working) q3_ui: added sv_punkbuster toggle in start server menu + automated building of the new PK3s, unix/Conscript-pk3 2002-7-25 Timothee Besset + added PB build scripts on Linux, fixed the Linux build 2002-7-12 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=511 fixing re.SetColor crash for widescreen displays (q3dm11) was calling to the renderer while not registered 2002-6-19 Timothee Besset + r_roundImagesDown 0 + map q3dm16 -> tr_image.c ResampleTexture crash buffer overflow because of resample to 2048x.. xian_q3dm12_leftwall4fin.jpg 1152x384 bumped one buffer byte p1[1024] -> byte p1[2048], added a safe check 2002-6-14 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=493 propagate a renderer fix from RTCW. fixes a one-frame visual glitch when mod code registers a shader after drawsurfaces are generated but before the frame is rendered 2002-6-12 Timothee Besset + added cons and pcons to unix/, updated the build script 2002-5-24 Timothee Besset towards a new Q3 release? some bug fixes require protocol change, or mod code/mod interface change to be fixed properly this is a biz decision, dunno yet if we are going to want a new protocol (probably not) -> have to create a branch for the 1.31b, i.e. backwards compatible with 1.31 'Stable-1_31' and put the 1.32 specific / protocol changes on trunk no telling what will go in SOS in the end .. probably 1.32 2002-5-5 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=491 adding a sv_lanForceRate (defaults to 1) to turn on/off server forcing rate of LAN clients (only affects LAN dedicated clients - dedicated 1, default behaviour forces LAN clients to 99999 rate) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=470 fixing potential overflows with cl_cdkey (propagated from RTCW) + cons-based build system (imported from Wolf, was partly written for mod tools release already) building with SMP on by default + better #ifdef SMP handling ('disabled at compile time' message) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=494 Q_vsnprintf for vsprintf calls in the core not putting this in game code as we'd need a vsnprintf implementation in bg_lib.c 2002-4-5 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=462 taking out the fix which was found broken and incomplete 2002-8-4 Timothee Besset + adding NO_MOUSEGRAB define (select in the Makefile) 2002-2-4 Timothee Besset + applying Gareth's SMP patch + count number of CPUs (Sys_ProcessorCount in unix_shared.c), default r_smp appropriately + bumping version to 1.32 + if XInitThreads fails, set r_smp to zero 2002-28-2 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=462 send potential remaining fragmented packets before sending a gamestate 2002-26-2 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=455 removed old libMesaVoodooGL.so loading code Voodoo cards should use XF4/DRI, that load code was outdated and confusing people with broken OpenGL 2002-16-1 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=441 adding brandelf calls to the setup building process so that our binaries run on BSD 2002-1-1 Timothee Besset + updated FAQ with BSD info (bug #441) + FAQ update on CLIENT_UNKNOWN_TO_AUTH + FAQ update for proper strace usage 2001-12-12 Timothee Besset + Q3 1.31 point release updating build_setup.sh to new pk3 files (baseq3/pak7.pk3 missionpack/pak2.pk3) + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=395 adding quake3.xpm icon, and modified the setup accordingly to put symlinks + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=390 ignoring SIGTTIN SIGTTOU 2001-06-12 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=402 bug with full scene 2001-04-12 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=398 cg_bobup cheat protect + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=399 fixed Setup > System > Driver info crash + checked in code/spank.sh script, perform checksuming 2001-18-09 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=371 propagating sound code fixes from Wolf to Q3 2001-11-08 Timothee Besset + setup script was still broken, damn shell expansion the exit code for Q3 was always zero instead of $? propagating the fix to Wolf 2001-11-04 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=382 modified challenge code for motd to be truly random 2001-10-31 Timothee Besset Moved updated q3asm and lcc source at the toplevel, MissionPack/q3asm and MissionPack/lcc 2001-10-29 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=381 build system is now functional 2001-10-21 Timothee Besset + updated Sys_LoadDll code on linux to search in the following order: #1 current directory #2 fs_homepath #3 fs_basepath this was needed to make mod developement easier 2001-10-09 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=51 the code to buffer the redirection was in there but disabled? (Com_Printf) enabled it back + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=52 connection issues / userinfo client side fix, instead of sending 'connect ' packet we now send 'connect ""' 2001-10-08 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=371 added a PROT_READ to the mmap call this was needed to go around a bug in glibc i586 i686, memset doing read access since the audio_fd is opened O_RDWD this is harmless to Q3 2001-10-07 Timothee Besset + updating from SOS S_WriteLinearBlastStereo16 C/asm is back in snd_mix.c (Graeme) r_showtris r_shownormal cheat protections + Sys_LoadDll changes: removing -debug search when loading native dlls changing the fatal aborts when not finding native from release only to debug only (was a misfeature) used to search in cd_path which is bogus, now searching in pwd if basepath fails + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=275 fixed r_fullbright not being cheat protected / was a CVAR_LATCHED|CVAR_CHEAT issue 2001-09-06 Timothee Besset + updated from SOS, some changes to qcommon/unzip.c (statics) 2001-08-27 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=3 Added some code in CL_InitDownloads to use FS_ComparePaks and print out information about server-referenced paks that are missing on the client. It is a first step, allows to get precise information about what can cause a connection to fail (typically when the user is sent back to the main screen). 2001-08-22 Timothee Besset + https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=86 fixed sound bug (with Graeme hints) 2001-08-20 Timothee Besset + made sure Sys_Printf doesn't get into an endless loop if logfile is on fixed qconsole.log issues, +set logfile 1 +set fs_debug 1 was crashing (any OS) fixed logfile 1 / ttycon 1 issue, didn't exit properly (same endless looping) also fixes an issue reported by q3f team + changed rcon commands from Com_DPrintf to Com_Printf so that they show up in the console (with IP information) 2001-08-19 Timothee Besset + fixed https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=91 (autodownload toggle in q3 ui) + fixed https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=76 g_password issue + fixed https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=93 cheat protecting r_lodCurveError + wontfix https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=92 2001-08-18 Timothee Besset + more fixes to the 7-button mouse code (linux only) + updated faq about gamma slider + added "servers don't show up in ingame browser" to faq + added Alt+Enter toggle for fullscreen/windowed (linux) 2001-08-16 Timothee Besset reconfiguring CVS repository to give access to Gareth + testin gareth's access 2001-08-03 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=85 fixes in the setup code for older bash versions 2001-08-02 Timothee Besset * commented out assembly implementation of S_WriteLinearBlastStereo16, using modified C implementation from Zaphod need to check performance: https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=88 * finished const declarations in CG_Trace calls, was needed in pmove_t declaration and some other functions cgame/cg_local.h : CG_trace trap_CM_BoxTrace game/bg_public.h : using const in pmove_t trace functions prototypes (fixes gcc warnings: assignment from incompatible pointer type) 2001-07-26 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=78 mapped K_MOUSE4 K_MOUSE5 2001-07-23 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=5 more fixes, handling meta characters and various kinds of backspace 2001-07-22 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=5 after testing feedback, fixed more stuff: better backspace, works with putty and potentially more terminals * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=51 band aid fix to rcon status, incresed MAX_PUSHED_EVENTS from 256 to 1024 (adds 28kb of mem requirements) 2001-07-21 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=2 using XF86 Gamma extension to set the gamma in game from the menus (previous behaviour was to set /r_gamma and restart, renderer relying on s_gammatable) restoring initial gamma on GLimp_ShutDown 2001-07-19 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=5 first usable version of dedicated console added history and completion functionality ready for some testing still some TODOs and FIXMEs: keep the currently edited line when going back from history exploration edit the current line with cursor, insert mode etc. 2001-07-18 Timothee Besset * starting TAB completion and history for the dedicated server (tty console) removed Sys_ConsoleOutput (unused) removing bogus nostdout variable cleanup of a big chunk of code that Bernd commented out and scheduled for deletion moved completion code from client/cl_keys.c stuff into qcommon/common.c, Field_CompleteCommand(field_t*) 2001-07-13 Timothee Besset * fixed https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=8 screenshots overwrites * fixed https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=31 DOUBLE SIGNAL FAULT 2001-07-11 Timothee Besset * fix for french keybards / console toggle / bound to XK_twosuperior 2001-07-10 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=19 cleanup of the keyboard code, adding com_developer message in case XLookupString would fail 2001-07-10 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=33 using our custom handlers for X errors, should make things more robust (X docs say some X errors are not fatal, but the default X handler exits the app anyway) 2001-07-08 Timothee Besset * https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=19 keyboard state issues, fixed the sticking with ctrl key (thks relnev) 2001-07-07 Timothee Besset * closing https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=13 the fixes to bug #9 solved this one too * checking in to SOS 2001-07-05 Timothee Besset * work on https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=9 filesystem code changes: updated the documentation in files.c to the current system added correct fs_homepath fs_basepath fs_cdpath scanning to FS_SV_FOpenFileRead (fixes description.txt not found, and probably a few other linux issues) 2001-06-29 Timothee Besset * fixed setup issues (graphical/console) https://zerowing.idsoftware.com/bugzilla/show_bug.cgi?id=6 2001-06-26 Timothee Besset * bug tracker is online at https://zerowing.idsoftware.com/bugzilla authentication, use login: bugs password: b00gies for now, using it as the linux bug tracker, possible use for more OSes and programs if anyone is interested. * tweaked the graphical setup to send to bugs@idsoftware.com on errors instead of support@lokigames.com 2001-06-19 Timothee Besset * fixed generated launch script /usr/local/bin/quake3, exit $* should be exit 2001-06-18 Timothee Besset * rebuilt 1.29f setups, released as 1.29f-beta1 'Q3 1.29f linux-i386 Jun 19 2001' 2001-06-10 Timothee Besset * rebuilt against PR source, packaged 1.29b setups 2001-05-25 Timothee Besset * graphical setup, based on Loki's setup tool (GPL) 2001-05-22 Timothee Besset * changed fs_basepath to fs_homepath, according to Graeme's changes (probably missed this change?) this fixes the q3key prompting at each game startup 2001-05-20 Timothee Besset * rebuilding 1.28b, various fixes on linux build: - SetProgramPath was renamed to Sys_SetDefaultCDPath in unix_shared.c updated unix_main.c accordingly - some prototypes in qgl.h are guarded by #ifndef GL_VERSION_1_2 (ARB extentions) those prototypes are needed by linux_glimp for importing functions and casting, added a #ifdef __linux__ (not a clean solution) - game/q_shared.h little endian / big endian functions have been added gcc generates warnings about functions being unused .. inlined them - cgame/cg_marks.c // TTimo // gcc warning: might be used uninitialized float sInc = 0.0; float s = 0.0; 2001-05-15 Timothee Besset * fixes to linux Makefile for bspc 2.1h * various updates to 1.28b on linux 2001-05-09 Timothee Besset * R. Duffy reverted game/bg_pmove.c PM_CheckDuck, was a merging screup on my side * updated setup to 1.27z, removed the .so from the setup distribution (they were in 1.27g because of issues) FIXME: gotta get pk3's first 2001-05-04 Timothee Besset * fixes to gcc, building RC for 1.27s 2001-05-01 Timothee Besset * added qcommon/huffman.c to the Makefile * gcc -Wall: commenting out CL_Netchan_Encode CL_Netchan_Decode (cl_net_chan.c) Netchan_ScramblePacket Netchan_UnScramblePacket (net_chan.c) SV_Netchan_Encode SV_Netchan_Decode (sv_net_chan.c) 2001-04-26 Timothee Besset * fixed dedicated server crash when entering the VM_COMPILED qagame on a mod (some statics lacked initialization) 2001-04-25 Timothee Besset * added $(Q3POBJ) to clean target (cleanup of platform-dependent objects) * more make clean improvements 2001-04-23 Timothee Besset * cleanup the mod selection code, remove duplicates * some issues with release builds, my main developement box doesn't build stable binaries with release settings removing -fomit-frame-pointer seems to fix (there's probably a performance hit) see OMIT-FRAME-POINTER.txt 2001-04-13 Timothee Besset * checked in a first set of merged files 2001-04-06 Timothee Besset * merged back the core linux parts to make 1.27g linux build from the Source Safe tree again 2001-02-27 Bernd Kreimeier * CVS: tag with changes as of today cvs tag id1-27j-loki01027 * code/qcommon/msg.c: numFields loop (SOS). * code/qcommon/files.c: ue Q_stricmp (SOS uses stricmp, was strcmp). * code/game/q_shared.h (Q3_VERSION): 1.27j. Also MAX_STRING_TOKENS upped from 256 to 1024 (SOS). * code/server/sv_snapshot.c (SV_AddEntitiesVisibleFromPoint): see below. * code/game/g_public.h (SVF_NOTSINGLECLIENT): added (SOS). * code/server/sv_ccmds.c: see below. * code/game/g_main.c: g_gametype cvar now userinfo (SOS). * code/game/g_active.c (SendPendingPredictableEvents): new (SOS). * code/game/bg_misc.c: new SOS (sos010227) * SOS: new update sos010227. 2001-02-22 Bernd Kreimeier * CVS: now in sync with last SOS and cleanup up cvs tag id1-27i-loki01022 * code/ui/ui_shared.c: below. * code/ui/ui_main.c: leftover code! * code/server/sv_world.c: below. * code/server/sv_snapshot.c: below. * code/server/sv_init.c: below. * code/server/sv_game.c: below. * code/server/sv_client.c: below. * code/server/sv_ccmds.c: below. * code/server/sv_bot.c: below. * code/server/server.h: below. * code/renderer/tr_surface.c: below. * code/renderer/tr_shader.c: changed assert to early return. * code/renderer/tr_shade_calc.c: below. * code/renderer/tr_shade.c: below. * code/renderer/tr_scene.c: below. * code/renderer/tr_mesh.c: below. * code/renderer/tr_local.h: below. * code/qcommon/vm_x86.c: cleanup. * code/qcommon/vm.c: below. * code/qcommon/unzip.c: below. * code/qcommon/qcommon.h: below. * code/qcommon/files.c: below. * code/qcommon/cvar.c: cleanup. 2001-02-21 Bernd Kreimeier * code/qcommon/common.c: cleanup. * code/qcommon/cm_trace.c: cleanup. * code/qcommon/cm_patch.c: cleanup. * code/qcommon/cm_public.h: cleanup. * code/game/q_shared.h: cleanup. * code/game/q_shared.c: cleanup. * code/game/q_math.c: cleanup. * code/game/g_syscalls.asm: changed (once more) floor,ceil etc. * code/game/g_spawn.c: cleanup. * code/game/g_session.c: cleanup. * code/game/g_cmds.c: cleanup. * code/game/g_client.c: cleanup. * code/game/g_arenas.c: cleanup. * code/game/bg_slidemove.c: cleanup. * code/game/bg_pmove.c (PM_CheckDuck): old call to trace? * code/game/bg_misc.c: cleanup. * code/game/be_aas.h: dead code. * code/game/ai_dmq3.c: cleanup. One clear/copy switched? * code/game/ai_dmnet.c: more //*/. Why oh why not DEBUG.... * code/client/snd_mix.c: below. * code/client/snd_dma.c: below. * code/client/keys.h: cleanup. TODO: #error in q3_ui/keycodes.h ? * code/client/client.h: cleanup. * code/client/cl_main.c: misplaced bracket. Cleanup. * code/client/cl_keys.c: below. * code/client/cl_cin.c: below. * code/client/cl_cgame.c: cleanup. TODO: define assert for Win32 or guard my assertions. * code/cgame/cg_syscalls.c: below. * code/cgame/cg_servercmds.c: below. * code/cgame/cg_players.c: cleanup. * code/cgame/cg_newdraw.c: remember to diff against cg_newDraw.c in SOS (mixed case). TODO: get id to use cg_newdraw.c, and to remove cg_newDraw.c/cpp. * code/cgame/cg_main.c: below. * code/cgame/cg_local.h: below. * code/cgame/cg_event.c: below. * code/cgame/cg_drawtools.c: below. * code/cgame/cg_draw.c: cleanup. * code/cgame/cg_consolecmds.c: dead code. * code/bspc/qbsp.h: below. * code/bspc/l_poly.c: below. * code/bspc/l_math.c: cleanup. * code/bspc/bspc.c: cleanup. * code/bspc/be_aas_bspc.c: cleanup. * code/bspc/aas_map.c: kept comments - merge loss at their end? * code/bspc/aas_file.c: cleanup. * code/botlib/be_interface.c: this file is plain impossible. There are layers of code made dead with /* */ and the resurrected by //* or // /* or variations of this. I reverted to exact mirror image of SOS to be sure - short of removing it's too easy to mistake live code for dead one. Later: have to change 5 occurences to avoid gcc complaints about nested comment tokens. TODO: somebody please get rid of the cruft in here. * code/botlib/be_ai_move.c: redundant typedef. * code/botlib/be_ai_chat.c: assertions on signed string index. Note: this is not in my ChangeLog - ouch. TODO: use gcc -fsigned-char on all platsforms to enforce Win32 TODO behavior (PPC has a default unsigned char, Intel has not). * code/botlib/be_aas_sample.c (AAS_TraceClientBBox): one code block was placed in different location, and one FPE hack not used. I would expect that divide by zero will still occur here. * code/botlib/be_aas_reach.c: below. * code/botlib/be_aas_cluster.c: cleanup. * CVS: the last tag (below) marks the version with a lot of history and additional comments. I am now bringing the codebase in sync with SOS as of yesterday, cleaning out comments, dead code and other differences to minimize a diff - in a valiant if futile attempt to roll back changes into the id codebase. Note: I ignore the $SOS$ - these are unfortunate but will change in the same awkward way at their end. Note: I stick to #if 0 instead of C comments around dead code id kept (nested comments issue). The commentary is changed to sosYYMMDD and includes the token DEAD. 2001-02-20 Bernd Kreimeier * CVS: update, then tag current version as cvs tag id1-27i-loki010219 * SOS: patched up to sos010219. * code/qcommon/cm_trace.c (CM_Trace): fabs on sphere offsets (SOS). * code/game/bg_slidemove.c (PM_StepSlideMove): stepSize vs. STEPSIZE (SOS). * code/game/bg_pmove.c (PM_CheckDuck): fix in stand up check (SOS). * code/bspc/bspc.c (main): -capsule (SOS). * code/bspc/qbsp.h: below (SOS). * code/bspc/be_aas_bspc.c (capsule_collision): added (SOS). * code/bspc/aas_map.c (CapsuleOriginDistanceFromPlane): added and used (SOS). * code/bspc/aas_file.c (AAS_WriteAASFile): removed diagnostics recently added. No matter how long you wait, they'll always get you ;-). * code/botlib/be_aas_cluster.c: enabled LogWrites, different flood (SOS). * SOS: patching up to snapshot sos010219. Note: For brevity, I use as marker sosYYMMDD now instead of bkYYMMDD, to distinguish from changes not in SOS. * CVS: tagged current version before patching up with SOS. cvs tag id1-27i-loki010216-bsd 2001-02-16 Bernd Kreimeier * code/server/sv_init.c: DLL_ONLY sets sv_pure to 0 and ROM. TODO: determine good sv_pure policy for DLL-only servers. * code/renderer/tr_shade_calc.c: my_ftol implementation (BSD). * code/unix/Makefile: FreeBSD sections. TODO: include target-specific Make-freebsd etc., include a Make-local not in CVS for build preferences, and generally clean up this mess. * code/unix/unix_glw.h: guard #error * code/unix/linux_snd.c: soundcard.h location (BSD). * code/unix/linux_glimp.c: guard system headers. Later: added Joystick stubs. Note: linux_ etc. prefixes start to loose meaning as we re-use most of this on UNIXes anyway. I didn't use Raf's freebsd_joystick.c but instead put generic stubs here. TODO: introduce generic -DNO_JOYSTICK flag. * code/renderer/tr_local.h: my_ftol guard. * code/renderer/qgl.h: FreeBSD guards. * code/qcommon/vm_x86.c: sys/types include on FreeBSD. * code/qcommon/md4.c: Win32 pragma guard. * code/qcommon/common.c: Com_Memcpy/Memset external. * code/game/q_shared.h: added FreeBSD defines. * code/game/q_math.c (BoxOnPlaneSide): FreeBSD conditional. TODO: check whether we have/need the assembly version anyway. * code/client/snd_mix.c: use C fallback on FreeBSD. Note: all of the above changes from the original port by Rafael Barrero. * CVS: tagged current version before merging FreeBSD related changes. cvs tag id1-27i-loki010215-ppc 2001-02-15 Bernd Kreimeier * code/unix/Makefile: BSD related changes. * code/cgame/cg_draw.c: hacked phone jack rendering check for Debug. TODO: finish Debug, fix CG_DrawDisconnect !!! * code/unix/vm_x86.c: error on compile attempts. Fight redundancy! * code/qcommon/vm_x86.c (VM_CallCompiled): dummy for linkage on PPC. Note: DLL_ONLY is the global Makefile option for DLL-only builts. Currently only executed on Linux. * code/unix/unix_main.c: *ppc postfix for DLLs. Ignored the changes to redundant code (have to remove the unused Un/LoadDll/API calls). * code/server/sv_game.c (VMA): changed macro (see below). PPC. * code/qcommon/vm.c (VM_DllSyscall): see lengthy commentary by Ryan. The existing VM code makes certain assumptions about the layout of varargs on the stack, which fall apart with call conventions that don't even put all parameters on the stack (gcc on PPC, register-rich). Using a dedicated memory area as our own stack. This should actually be the default behavior. Later: make vm_* cvars INIT/ROM for DLL_ONLY target. * code/qcommon/common.c: PPC change (from Ryan Gordon). 2001-02-07 Bernd Kreimeier * code/unix/unix_main.c: disabled FPE for debug for the time being (that is, until I can figure out Program received signal SIGFPE, Arithmetic exception. RB_BeginSurface (shader=0x449572e0, fogNum=0) at ..//renderer/tr_shade.c:307 307 tess.shaderTime = backEnd.refdef.floatTime - tess.shader->timeOffset; without any NaN's involved. TODO: unmask other FPE's selectively (see Mike's Tribes2, no getenv though). 2001-02-06 Bernd Kreimeier * SOS: up to date with todays snapshot. Note: got the date wrong, comment used was bk010205. Duh. * code/server/sv_snapshot.c (SV_UpdateServerCommandsToClient): below. * code/server/sv_main.c (SV_ReplacePendingServerCommands): new (SOS). * code/server/server.h: reliableSent (SOS). * code/renderer/tr_shade.c (ProjectDlightTexture): see below. * code/renderer/tr_scene.c: see below. * code/renderer/tr_public.h: see below (SOS). * code/renderer/tr_local.h: additive light support (SOS). * code/qcommon/cm_trace.c (CM_Trace): new tw.sphere.use branch (SOS). * code/game/g_spawn.c: notta, notq3a entities (SOS). * code/game/ai_dmq3.c: MAX_ACTIVATEAREAS search (SOS). * code/client/cl_cgame.c: see below. * code/cgame/cg_syscalls.c (trap_R_AddAdditiveLightToScene): below. * code/cgame/cg_syscalls.asm: see below (trap_R_AddAdditiveLightToScene). * code/cgame/cg_public.h: CG_R_ADDADDITIVELIGHTTOSCENE (SOS). * code/bspc/l_math.c: new VectorLengthSquared, removed rotate/matrix (SOS). * code/bspc/bspc.c (BSPC_VERSION): was 2.1e, now? * code/bspc/be_aas_bspc.c (BotImport_Trace): CM_BoxTrace sig. (SOS). * code/bspc/aas_file.c (AAS_WriteAASFile): SOS. * code/botlib/be_aas_sample.c (AAS_DeAllocAASLink): SOS. * code/unix/unix_main.c (Sys_LoadDll): do not load from installdir in NDEBUG (confusing relic from old Makefile). Postfix -debug.so for debug binaries to let both builds coexist. * code/unix/Makefile: updated install targets and VERSION. * Win32: build from SOS snapshot. Note: Unix CR/LF in *.dsw/*.dsp fucks up MSVC++. 2001-02-02 Bernd Kreimeier * SOS: all changes up to today. * code/server/sv_init.c (SV_TouchCGame): added. Also memset on reallocated client data (SOS). * code/qcommon/qcommon.h: see below. * code/qcommon/cvar.c (Cvar_SetLatched): new (SOS). * code/qcommon/cm_trace.c: more sphere test fixes (SOS). Note: SOS encryption key expired and updated by MrElusive. * code/qcommon/cm_patch.c (CM_TraceThroughPatchCollide): fix from MrElusive, fall through curved corner floors (q3dm17). Later: also in SOS (so is shadow FPE fix). * Win32: can't get an unadulterated SOS snapshot to build. First, fix CR/LF back again (Linux client converts all). find . -name '*.ds*' -print alias dos2unix='recode ibmpc..lat1' alias unix2dos='recode lat1..ibmpc' Next, find a *.dws that works? Nope, no cigar. 2001-02-01 Bernd Kreimeier * Win32: have to update dsp/dsw/etc. files in CVS, too. * CVS: tag previous version before update cvs tag id1-27h-loki010131-beta3 * code/game/q_math.c (Q_rsqrt): guard, #ifndef __linux__ for assert (for Win32 build). TODO: assert replacement for Win32? * code/q3_ui/ui_qmenu.c: see below. * code/q3_ui/ui_players.c: see below. * code/q3_ui/ui_controls2.c: float const with f postfix Note: Win32 C4305 warning. Somebody at id has been doing a lot of these recently as well... * code/cgame/cg_players.c (CG_PlayerShadow): applied fix by MrElusive, removed FPE hack (player shadows on zero mormals). Prolly in this evenings' CVS. * code/server/sv_game.c: new signatures (capsule again). * code/server/server.h: new signatures (SV_Trace,ClipToEntity). * code/server/sv_bot.c: new signatures (above). * code/qcommon/cm_trace.c: a truckload of changes. Math code added before moved upwards. Capsule traces added all over the place, old box traces moved in conditional branches, functions renamed and wrapped. Eliminated some of the previous' versions deadcode to keep diffs smaller. TODO: once a point release is out and reasonably bug TODO free, remove // bkYYMMDD annotations where SOS related. * code/qcommon/cm_public.h: new signatures in prototypes. * code/qcommon/cm_patch.c: dead code re-enabled, new sections (conditional branches for spheres) added to several trace functions. * code/qcommon/cm_local.h (CAPSULE_MODEL_HANDLE): added. * code/qcommon/cm_load.c (CM_TempBoxModel): capsules. * code/game/q_shared.h (Q3_VERSION): 1.27i now (new QVM traps). * code/game/g_syscalls.asm: see below. * code/game/g_public.h (SVF_CAPSULE): added (SOS). Also G_ entry poiints for capsule traces. * code/client/cl_cgame.c: see below. * code/cgame/cg_syscalls.c: see below. * code/cgame/cg_syscalls.asm: see below. * code/cgame/cg_public.h: new capsule trace code (SOS). 2001-01-31 Bernd Kreimeier * Win32: test compile (WinCVS, MSVC++). Have to guard isnan. Note: too much shit going on.... 2001-01-30 Bernd Kreimeier * CVS: update for patching up (pre-1.27i). * SOS: new changes (new collision detection primitives). Now Version 1.27i. TODO: start testing using DLL's (QVM code is out of sync). 2001-01-25 Bernd Kreimeier * SOS: caught up till today (below). * code/qcommon/cm_trace.c: new functions added: RotatePoint, TransposeMatrix, CreateRotationMatrix (SOS). (CM_TransformedBoxTrace): new rotation code used here. * code/q3_ui/ui_demo2.c: sizeof(extension). SOS. * code/game/g_cmds.c (G_SayTo): CON_CONNECTED. * code/game/ai_main.c: HOOK added (SOS). * code/botlib/be_aas_move.c (AAS_HorizontalVelocityForJump): correct fix for FPE occuring (SOS). * code/game/ai_dmq3.c: initmove.viewoffset (SOS). * code/game/q_math.c: guard asser/isnan with Q3_VM (q3asm). TODO: define Com_Error based assert macro? NDEBUG? 2001-01-24 Bernd Kreimeier * code/server/sv_ccmds.c (SV_MapRestart_f): some debug. TODO: map_restart 0 disconnects external client in 1.27h? * code/renderer/tr_image.c (LoadTGA): added some commentary and dead code based on fixes from GtkRadiant (Leonardo found flipped TGA's). 2001-01-23 Bernd Kreimeier * BETA3: finished testing, ready to upload to id FTP. Later: neither the FreeBSD beta not the Linux Beta3 uloaded. Beta2 not yet released, and clients get disconnected with Beta2 and Beta3 on SV_MapRestart_f. 2001-01-22 Bernd Kreimeier * code/client/cl_main.c (CL_InitDownloads): undid yesterday (SOS). * code/botlib/be_aas_sample.c (AAS_DeAllocAASLink): guard print (SOS). * code/server/sv_client.c (SV_DirectConnect): VM_Call disconnect (SOS). * code/qcommon/files.c (FS_ListFilteredFiles): trailing slashes (SOS). * code/game/g_cmds.c (SetTeam): print change (SOS). Note: the above plus VectorClear(v1) (below) are todays SOS changes. * code/cgame/cg_players.c (CG_PlayerShadow): ignore bogus (all zero) planes. This caused FPE in ProjectPointOnPlane. TODO: why does trace return zero normal planes? Note: gdb seems totally at loss with vec3_t arrays.... * code/botlib/be_aas_sample.c (AAS_TraceAreas): FPE. NaN in uninitialized v1 that wasn't supposed to be referred to in this branch. * code/botlib/be_aas_move.c (AAS_HorizontalVelocityForJump): FPE divide by zero (zero zvel, zero t) for jump estimates. * code/client/cl_main.c (CL_Frame):1856. uivm==NULL on client after server crashed. TODO: check that uivm always non-NULL for client. TODO: do setenv(FX_NO_SIGNALS) to avoid exit errors... * code/unix/linux_glimp.c (GLW_SetMode): added "Indirect" Mesa token to software rendering detection. Reworded error output and added drivername. TODO: measure framerate instead? 2001-01-21 Bernd Kreimeier * SOS: caught up with changes up until today. * code/server/sv_init.c (SV_SetConfigstring): gentity != NULL * code/server/sv_client.c: connect to "{all bots" server. * code/renderer/tr_init.c: JPEG extension on screenshots * code/qcommon/files.c: modes based on mods, fs_basegame * code/q3_ui/ui_demo2.c: dm3 extension (demo names, protocol). * code/game/g_client.c: savedEvents[] removed. * code/game/bg_misc.c: event sequence fixes. * code/client/snd_dma.c (S_StopBackgroundTrack): different use. * code/client/cl_main.c: demo file handling changed (names). Also CL_InitDownloads: always next download. * code/cgame/cg_servercmds.c: cg_thirdPerson. * code/cgame/cg_weapons.c: see below. Also CG_ShotgunPattern: different call (seed parameter). * code/cgame/cg_main.c: see below. * code/cgame/cg_local.h: new cg_noProjectileTrail Cvar. * code/cgame/cg_effects.c (CG_BubbleTrail): early out (above). * code/bspc/l_poly.c (BOGUS_RANGE): increased. * code/bspc/bspc.c: applied patch up to "2.1e" 2001-01-18 Bernd Kreimeier * code/ui/ui_main.c: below. * code/q3_ui/ui_main.c: UI_HASUNIQUECDKEY comment. Note: mods have to return qfalse. See Bug #2890 in Fenris. 2001-01-17 Bernd Kreimeier * BETA2: finished testing, uploaded to id's FTP for release. 2001-01-16 Bernd Kreimeier * CVS: checking in preparation for Beta2. cvs tag id1-27h-loki010116-beta2 * SOS: new bspc "2.1e". No change on 1.27h. * TEST: patch-up seems to work fine. No new files have been added to the linkage (i.e. the ft2/ files now added), so we might not be feature complete. * code/game/g_active.c (ClientThink_real): id MISSIONPACK conditional in addition to the ones I added earlier. * code/qcommon/files.c: REJECT. Linux hack for userdir threw it off. * code/qcommon/unzip.c: REJECT. CRC-32 section removed. Later: unused tempB * code/q3_ui/ui_syscalls.asm: REJECT. Start/StopBackgroundTrack. * code/ui/ui_syscalls.asm: REJECT. syscalls ids from 1.27h as of SOS (floor/ceil - will this ever get straightened out) * code/win32/win_input.c: REJECT. g_pMouse edit. * ui/menus.txt: REJECT. Replaced with 1.27h version. Note: some more due to $SOS$. * ui/: new scripts. cinematicmenu.menu, demo_quit.menu, ingame.txt, serverinfo_old.menu vid_restart.menu * code/ft2/ttconfig.h: below. * code/ft2/sfconfig.h: below. * code/ft2/pstables.h: below. * code/ft2/psnames.c: below. * code/ft2/psdriver.h/c: below. * code/ft2/keys.h: below. * code/ft2/ftbbox.c: new in 1.27h * code/cgame/cg_newdraw.c: beware: cg_newDraw.c gets lost in diff easily. * code/cgame/cg_rankings.c: file removed from SOS. 2001-01-15 Bernd Kreimeier * Patch-up: patching up from RC4 to 1.27h current. No changes since 010112 snapshot. ln -s sos010112/ work diff -urbB sos001204-rc4/ work > work.diff ln -s cvs-1.27g/ work patch -p0 < work.diff > work.patch find cvs1.27g/ -name '*.rej' -print * SOS: adding the remaining SOS snapshots to CVS. cvs import Quake3_sos sos001211 pr1-27g-win32-001211 Note: at this point id warned about repository corruption. Watch out for the syscall stuff in particular. cvs import Quake3_sos sos010104 pr1-27g-win32-010104 cvs import Quake3_sos sos010108 pr1-27h-win32-010108 cvs import Quake3_sos sos010110 pr1-27h-win32-010110 cvs import Quake3_sos sos010112 pr1-27h-win32-010112 Note: the first 1.27h might be the public (server only) beta released, the second one was post release. Beware of source files added and removed (botlib headers, FT2). Note: why so late? Don't ask... 2001-01-08 Bernd Kreimeier * SOS: id's working up to 1.27h (server side fix for Guard exploit seems to force earlier release). Updating CVS (most of the changes are debug code put in and then disabled, plus some fixes as below). Next patching up to current SOS. 2001-01-07 Bernd Kreimeier * Makefile: need to rework this for multiple platforms. We also need null/null_vm.c for platforms where we don't have JIT (assembly emit). 2001-01-04 Bernd Kreimeier * code/q3_ui/ui_connect.c (UI_DisplayDownloadInfo): time information for current (vs. start of download) is wrong, thus negative 1 "estimated time", as well as transfer rate just negative downloadSize. Not fixed. * code/unix/unix_main.c (Sys_ParseArgs): added. Note: for support/us, to identify builts. This is only a skeleton right now - if I ever feel the need to support more than "-v" and "--version" I'll have to flesh this out. * code/unix/linux_glimp.c (signal_handler): see below. * code/unix/unix_main.c (Sys_Exit): added an abstraction layer for exit/_exit/assert/raise issues. Note: need both a better debug/backtrace handling, and have to find a way to determine why/where the alleged startup/exit errors happen... 2001-01-03 Bernd Kreimeier * code/game/g_mem.c (G_Alloc): ERR_DROP initiated by addbot commands for large sv_maxclients, allegedly caused segfaults in 1.17. Not reproducible. TODO: recover more gracefully from failure to add bot? * code/renderer/tr_light.c (R_LightForPoint): Tim Angus reports a crashbug with nolight maps. Also assertion in R_SetupEntityLightingGrid, might want conditional there. DONE: fixed crash on LightForPoint for nolight maps. * code/qcommon/qcommon.h: NUM_SERVER_PORTS. A feature request to increase this, or make it more flexible otherwise (Fenris). TODO: id decision on more flexible NUM_SERVER_PORTS. 2001-01-02 Bernd Kreimeier * code/unix/snapvector.nasm: fixed FPU bit (the current one had reserved bits off, behavior should not change). * code/qcommon/vm_x86.c: fixed symbols (below). * code/unix/ftol.nasm: FPU bits weren't correct (duh). DONE: shoot-though floor (q3dm5) DONE: cursor-in-rect off (TA/Player model selection) Note: in gdb, "disassemble " is your friend. * code/cgame/cg_public.h: CG_MEMSET is set to 100. In cg_syscalls.asm it's 101. If I change it I get Bad trap 100 from the cgame VM code, so the 1.27g "official" VM code uses it. * code/unix/linux_common.c: have to fall back to C, the current assembly is buggy... * code/unix/Makefile (linux_common.o): added. Later: also for dedicated. Less portable this way. TODO: C_ONLY for dedicated on non-i386 only? * code/qcommon/common.c: do not use memcpy/memset under Linux. * code/unix/linux_common.c: added Andrew's assembly port. TODO: C_ONLY for Com_Memset/Memcpy? Conditionals are fubared. * code/qcommon/vm.c (VM_Init): use Win32 defaults (do not use DLL's by default). This exposes DLL rounding errors (damage through floors), and we don't want DLL's used by default anyway. TODO: why vm_ui default of 1? * code/botlib/l_precomp.c (SourceWarning): removed assert. * code/game/bg_lib.c (acos): defined, but we don't actually use it except where the cg_syscalls.asm trap is used. * code/game/g_public.h: missing lots of trap tokens. * code/game/g_syscalls.c: missing lots of traps. * code/game/g_syscalls.asm: more inconsistent hooks, were: equ floor -111 equ ceil -112 equ testPrintInt -113 equ testPrintFloat -114 now changed to match cg_syscalls. Note: fixed this in UI earlier, how did this slip through the diffs against SOS? * code/game/g_syscalls.c: no acos hook. * code/cgame/cg_syscalls.c: no acos hook. * code/cgame/cg_syscalls.asm: has acos hook as -112 Note: report from Tim Angus. The acos function is in bg_lib.c which is linked only into ui (not q3_ui). That means we are using libc acos right now? Note: QVM traps are negative? * BSD/Irix: tagged current CVS (not all of the below) as cvs tag id1-27g-loki010102-bsd1 for BSD work (Rafael Barrero). Also be used for Irix update. 2001-01-01 Bernd Kreimeier * SOS: adding the remaining SOS snapshots to CVS. cvs import Quake3_sos sos001201-rc3 pr1-27f-win32-001201-rc3 cvs import Quake3_sos sos001202 pr1-27f-win32-001202 cvs import Quake3_sos sos001204 pr1-27g-win32-001204-rc4 This is the codebase to which the Linux branch has been patched up. I can't verify whether this is identical to the RC4 codebase as the tag doesn't work (but can check against the ZIP file..) cvs import Quake3_sos sos001211 pr1-27g-win32-001211 The above snapshot contains a (post-release?) fix to ui_syscalls in ui/ and q3_ui/. This change has been used in Linux (Beta1 and above). At this point, id discouraged further use of SOS due to repository corruption on their end. No further snapshots were taken since. * Fenris: since the release of the Beta1 bugs have been maintained at http://fenris.lokigames.com/. I am going to list issues here as they get fixed. 2000-12-21 Bernd Kreimeier * code/renderer/tr_font.c: graceful silence with old mods? * code/botlib/l_precomp.c (SourceWarning): graceful exit if old mod? 2000-12-20 Bernd Kreimeier * code/server/sv_ccmds.c (SV_MapRestart_f): see below. * code/qcommon/vm.c: currentVM is 0x0 in VM_ArgPtr. In VM_Call, oldVM was NULL - made conditional the reset of currentVM to oldVM. 2000-12-18 Bernd Kreimeier * BETA1: closed Linux beta release. Stripped debug and release binaries, DLL's, and pak4.pk3. CVS checkin, will be tagged as cvs tag id1-27g-loki001218-beta1 Later: id added a pak5.pk3 to the Win32 point release, added this to the BETA1 best. * code/qcommon/vm_x86.c: C37F. * code/unix/snapvector.nasm: C37F. Note: short of any real evidence, I gamble and use max. precision (as well as default Linux precision, but NOT Win32 precision). It seems that precision change is not really an issue (despite Graeme's claim that the cursor in the menu was/is off). I also pick the roundiung behavior that is seemingly used by ANSI and gcc (but possibly not Win32 _ftol depending on build). 2000-12-15 Bernd Kreimeier * code/unix/Makefile: added snapvector.o * code/unix/unix_shared.c: #if 0'ed the old snapvector code. * code/unix/snapvector.nasm (Sys_SnapVectorCW): two new assembly functions from AndrewH that explicitely set the FPU control word to convert vec3_t, to ensure cross-platform behavior for both DLL and QVM. * code/unix/ftol.nasm (Q_ftolC37F): for globals. * code/unix/unix_main.c: took out global FPU manipulation. For clarity this should be VM only. * code/qcommon/vm_x86.c: added prototypes for the ftol library. To select a specific behavior for the entire VM, set ftolPtr accordingly. Later: the GCC ftol function of course affect the stack (there is no "declspec naked"). The problem seems to be that the VM never handles the stack in a way compatible to regular gcc C functions. For some odd reason _ftol seems to do the right thing under Win32. All 4 control words implemented at the moment work just fine with the menus. * code/unix/ftol.nasm: added a small library of "safe" qftol variations that explicitely set the control word to the relevant (4) possibilities. 2000-12-13 Bernd Kreimeier * code/qcommon/vm_x86.c: an entire day spent trying to nail the ftol issues. It breaks down like this: id used to use an unsafe (no setting FPU control word) fistp. That seemingly caused subtle physics bugs which nobody cared about in 1.17. They then changed the UI code, and ran into the UI bugs: menu entries shifted to the right, fonts vanishing. Then they switched to using _ftol. Then they had to reproduce the old behavior for the physics code due to public outrage. My original port used a simple (long)float cast, which gcc seemingly compiles to code that does OR 0C00 on whatever current control word (precision unchanged). This breaks the menus. If I use the unprotected fistp instead, which should (Linux 037F default) use "nearest/even", then my menus are correct. That would mean Win32 _ftol in id's compile does the same, only that would require /qifist or some equivalent compile flag, which I can't find. Two disassemblies of _ftol I got from others showed OR 0C00 as part of the default (ANSI) behavior. 2000-12-13 Bernd Kreimeier * code/game/bg_pmove.c (PmoveSingle): trap_SnapVector. The one true and single call to snap velocity. Note: bspc/map.c:void SnapVector(vec3_t normal) qcommon/cm_patch.c:void CM_SnapVector(vec3_t normal) game/q_shared.h: #define SnapVector(v) {v[0]=((int)(v[0]));... * code/client/cl_cgame.c: CG_SNAPVECTOR. * code/server/sv_game.c: G_SNAPVECTOR. Note: these go through trap_SnapVector in syscalls. * code/unix/unix_shared.c (Sys_SnapVector): sticking to old Linux version for now... * code/win32/win_shared.c (Sys_SnapVector): changed. Note: Graeme points out this was changed to fix ftol artifacts? TODO: calculate errors for various ftol variants... * code/qcommon/vm_x86.c: both the old fistp code (1.17) and the new qftol function apparatently work. Using the ftol.nasm code for now. * code/unix/Makefile: DO_NASM and ftol.o. * code/unix/ftol.nasm (qftol): created from Mike's SoF replacements, with Andrew's help to satify the VM stack/call requirements. TODO: use Q_ftol herein to replace myftol elsewhere. * code/unix/unix_main.c (Sys_ConfigureFPU): SIGFPE. TODO: divide by zero in botlib. Disable this for now. Note: we can't introduce calculation differences between versions, so fixing these will have to wait. * code/qcommon/vm_x86.c: two new lines in Win32 branch missing from Linux assembly in AsmCall: mov eax, dword ptr [edi] and eax, [callMask] Added, doesn't seem to affect UI etc. bugs. Later: no FTOL_PTR, use fistp non-IEEE assembly as in old version. This seems to work for Q3 and TA, while qftol (simple cast) does not - for Win32 Graeme says the reverse is true. * code/qcommon/vm_x86_old.c: used the old cvs-1.17 version. Two fixes (Hunk_Alloc, Com_Memcpy), and it works: +set vm_game 2 +set vm_ui 2 +set vm_cgame 2 UI, cgame and game w/o apparent problems. 2000-12-12 Bernd Kreimeier * code/unix/Makefile: cleanup of redundant flags. Removed bogus MALLOC_CHECK (note to self: export MALLOC_CHECK_=2). Also DO_SHLIB_CC on all UI DLL's. Added and removed DEBUG_VM flag. TODO: figure out whether Zoid did UI this way intentionally. Note: this seemingly fixed the botimport problem, although most of the changes were just redundant CFLAGS removed. Given our wanker toolchain, should have been more paranoid. All DLL's can now be used w/o apparent problems. * code/server/sv_main.c: gvm init. * code/server/sv_game.c: gvm assertions. * code/unix/unix_main.c (Sys_LoadDll): print vmMain Note: top no avail. There is some odd ld/gdb problem here that prevents examining globals and obfuscates part of the stack between VM_Call and lower level code, through G_InitGame. This is not just DLL's being loaded and unloaded. Wromg flags during build? The vmCvar for "bot_developer" ends up overlapping global botimport in memory, which thus zero-fills part of the function pointer table. * code/server/sv_bot.c (SV_BotInitBotLib): this (by way of GetBotLibAPI) is responsible for setting botimport, which, if using the game DLL, is not properly set up. Called in SV_Init(). * code/game/q_shared.c: Q_strncpyz does zero padding (duh). Note: calls strncpy, which does a zero fill up to destsize. If destsize exceeds memory size, zero padding will overwrite adjacent memory. Suspicion was this happend to botimport. * code/qcommon/cvar.c: possible problem in Q_strncpyz call. * code/botlib/be_ai_weap.c (weaponinfo_fields): made this static. Note: it seems that the "number" string got replaced by p def.fields[0] {name = 0x40000000 "\177ELF\001\001\001", offset = 2, type = 50, .. Memory corruption? * code/game/inv.h (WEAPONINDEX_GAUNTLET): defined here. * botfiles/weapons.c (Gauntlet): the baseq3/qagamei386.so parser breaks here: number WEAPONINDEX_GAUNTLET * code/botlib/l_precomp.c (SourceWarning): added assertion to trap botlib parsing problem.. * RC1: for beta test. Using my own vm/ui.qvm files in this case. TODO: Setup with nouninstall. TODO: fix game DLL/ botlib setup problem (so all DLL's work) TODO: SIGFPE TODO: profile? * code/unix/Makefile (ai_vcmd.o): added to game DLL linkage. How the fuck did this happen? DONE: "qagamei386.so: undefined symbol: BotVoiceChat_Defend" * TEST: +set vm_ui 2 (vm_x86, not interpreter). Breaks! Further: qagame had undefined, but seemingly gets reloaded second try (I hate the Linux linker). * TODO: never reload fail DLL, abort engine 2000-12-11 Bernd Kreimeier * TEST: recompile QVM/DLL and executable to test new UI code. The UI QVMs from the paks still do not work. * SOS: changes in UI code! * code/q3_ui/ui_public.h: this file is deprecated Note: e.g. it does not contain the background track calls. * code/ui/ui_public.h: the uiImport_t enum here determines the values. * code/ui/ui_syscalls.asm: same as q3_ui now, were: equ floor -111 equ ceil -112 * code/q3_ui/ui_syscalls.asm: these are now switched, were: equ trap_S_StartBackgroundTrack -63 equ trap_S_StopBackgroundTrack -64 The new values match the ui/ equivalent. Also, floor (-108) and ceil (-109) are different in ui/. * CVS: going to check in this snapshot and tag it as cvs tag id1-27g-loki001209-rc4 Presumed equivalent to SOS tag "1.27g RC4" (master). As I can't obtain the tagged code using SOS (neither Win32 nor Linux client) I can only guess. * TEST: use my own VM code, ion baseq3/vm/*.qvm and missionpack/vm/. This works - in other words, the menu bug seems in the UI code, and is fixed in my codebase. * TEST: make release. Note: I can postpone DLL specific problems. Bad performance is not as important as outright bugs. Thus the UI QVM issue is the only one that stops me from creating an RC. TODO: Q3 UI QVM code from pak file does not work (neither does TA). TODO: sound with video playback still awful. Threaded sound, I guess. TODO: ERROR: couldn't open demos/DEMO002.dm3.dm_48 (same demo001.dm3.dm_48) TODO: do not show Q3 demos in TA menu? TODO: new demos for Q3? Or at least error message? * code/game/bg_lib.c: itrinsics excluded by Q3_VM (another -O compile). Uninitialized variable. * code/unix/Makefile: -O for uninit on patched code. Also shortcuts. TODO: DC_ONLY seems an obsolete flag, used in Makefiles, not source. * TEST: +set sv_pure 0 +set vm_game 1 +set vm_cgame 1 +set vm_ui 0 Turns out that the pak0.pk3 UI QVM code is seemingly broken in TA and Q3, but my UI DLL is not. In reverse, the QVM game/cgame for Q3 seems to work quite well (including bots). The TA game/cgame also works, including botlib init. TODO: BotLib Init using game DLL gives: TODO: Error: file weapons.c, line 38: unknown structure field number TODO: Fatal: couldn't load the weapon config TODO: Error: BotLoadMap: bot library used before being setup * TEST: checked the rc4winstlr.zip CD tree against my test install. baseq3/pak4.pl3 and missionpack/pak0.pk3 are identical, but I finally recognized that there was a missionpack/pak1.pk3 not in the final install - left over from an earlier update from id. Doesn't seem to affect the DLL based runs at all. Note: I still do not have the final CD snapshot Robert promised me mid last week, they haven't even fixed the FTP account they took down. Communication with id is as abyssmal as ever. 2000-12-08 Bernd Kreimeier * TEST: running with RC4 data files. TODO: "bot library used before setup" (Q3+TA) TODO: Q3 old mods wreak havoc (graceful bounce) TODO: supress "FreeType code not available" in renderer TODO: can't move in Q3 TODO: items flicker in Q3 TODO: no decals in Q3 TODO: VM UI code still broken (Q3+TA) TODO: sound code is awful TODO: video playback inferior to earlier builds * code/q3_ui/ui_local.h: prototype trap_VerifyCDKey(..) * code/game/g_active.c ( StuckInOtherClient): TA only. * code/cgame/cg_draw.c: 4x unbalanced `#endif' - from patch? * code/null/null_client.c (CL_CDKeyValidate): dummy added. * code/qcommon/common.c: Q_acos missing, changed conditionals * code/qcommon/vm_x86.c: unreacheable _asm instruction that gcc doesn't quite like... #if 0'ed for now TODO: understand _asm { mov eax,[ebx] }, fix it for gcc * TEST: compile... * code/ui/ui_main.c: full REJECT. Manual merge. Note: preserved debug_protocol lines, who knows what it's good for. * code/qcommon/files.c: REJECT. SafeMode, demo server FS_Restart. * code/client/snd_mem.c: REJECT: $SOS$. * code/client/snd_dma.c: REJECT: $SOS$. * code/client/cl_cin.c: REJECT. com_timescale, $SOS$. * code/cgame/cg_draw.c: REJECT. Lots, but virtually all either float postfix (on some, not all places), or #ifndef MISSIONPACK that I had already put in during -Werror (conditional unused). * code/cgame/cg_consolecmds.c: REJECT. id commented unused code that I had #if 0'ed earlier. * code/game/: three REJECT for $SOS$. * code/botlib/: lots REJECT for $SOS$. * Patch: patching up from demo source. ln -s sos001204-rc4 work diff -urbB sos001122-demo/ work > work.diff ln -s cvs-1.27b/ work patch -p0 < work.diff > work.patch find cvs1.27b/ -name '*.rej' -print * CVS: going to check in this snapshot and tag it as cvs tag id1-27b-loki001208-demo Then patching up to RC4, as of sos001204-rc4 (no changes since, should be equivalent to SOS tag "1.27g RC4" (raduffy), i.e. master. * TEST: installed demota/ from Win32 distribution. Binary fails claiming "Corrupted pak0.pk3". Abandoned. Note: a Linux demo for Q3TA has no priority. Most important is the Q3A point release in time for Q3TA hitting shelves, followed by testing for Q3TA. The source is in CVS and tagged (see above) in case a demo matching the released files has to be provided later. 2000-12-07 Bernd Kreimeier * TEST: compile and link - succeeds. * code/ui/ui_main.c: UI_StopServerRefresh now uaws. New unused variables. * code/unix/unix_main.c: added Sys_LowPhysicalMemory() stub. TODO: write Linux equivalent to GlobalMemoryStatus. * code/qcommon/common.c: Com_Memset/Com_Memcpy. Neither assembly nor C versions included if not on Win32 i386. TODO: using/porting assembly? * code/qcommon/files.c: unused variable. TODO: fs_scrambledProductId unused if 0 for now. Note: -DFS_MISSING for id's pak cleanup, not used. * TEST: compile and link - fails. * code/macosx/Client/Makefile.postamble: empty ORIG. * code/macosx/Client/Makefile.preamble: ORIG. $(BOTLIB_OBJS) added. * code/server/sv_client.c: ORIG. Com_Memset. * code/renderer/tr_shader.c: ORIG. Com_Memset, CIN_Shader. * code/qcommon/vm_x86.c: ORIG. Com_Memcpy. * code/qcommon/unzip.c: REJECT. Com_Memcpy, $SOS$. * code/qcommon/qcommon.h: ORIG. PROTOCOL 47, plus Sys_LowPhysicalMemory. * code/qcommon/md4.c: Com_Memset,Com_Memcpy (ORIG). * code/qcommon/files.c (Sys_ConcatenateFileList): REJECT. Our additons threw it off, plus $SOS$. * code/qcommon/common.c: they fixed same unused variables (REJECT). * code/ui/ui_shared.c: additions (ORIG). * code/ui/ui_gameinfo.c: COM_Compress added (ORIG). * code/ui/ui_atoms.c: print statements removed (ORIG). * code/ui/ui_main.c (UI_DoServerRefresh): REJECT on comment edit... * code/game/g_cmds.c (Cmd_VoiceTaunt_f): logic changed heavily. ORIG. * code/game/q_shared.h: Q3_VERSION "Q3 Team Arena Demo 1.27b" plus Com_Memset, Com_Memcpy, CIN_shader, COM_Compress. * code/game/g_main.c: Cvar change only * code/game/ai_dmq3.c: $SOS$. * code/client/snd_mix.c: Com_Memset * code/client/client.h: additions (ORIG). * code/client/snd_mem.c: see below. * code/client/snd_dma.c: $SOS$ (CVS keyword). * code/client/cl_cin.c: they removed unused (REJECT). * code/cgame/cg_servercmds.c: ORIG. compress, noTaunt etc. * code/cgame/cg_main.c: ORIG. Conditonal branch, COM_Compress. * code/cgame/cg_consolecmds.c: ORIG. Cvar values changed. * code/cgame/cg_draw.c (CG_DrawTeamBackground): ORIG. no reject but *.orig file created. I just mark spots were code changed after verifying the patch succeeded. * code/cgame/cg_event.c: fixed reject (REJECT). * code/botlib/: all *.rej here due to SOS/CVS $Keyword$. TODO: preserve SOS comments/rev history somehow. * Patch: patching up to demo source. ln -s sos001122-demo work diff -urbB sos001119/ work > work.diff ln -s cvs-1.26/ work patch -p0 < work.diff > work.patch find cvs1.26/ -name '*.rej' -print * CVS: going to check in this snapshot and tag it as cvs tag id1-26w-loki001207 to prepare for upgrading to RC4. I have already made many more changes than I wanted to w/o getting any closer to pinpointing the problem, I might as well patch up to id's more current sources. * code/botlib/be_interface.c: initialize by memset. Turns out that this fails in Export_BotLibSetup on BotSetupWeaponAI loading "weapons.c" (from the pak, presumably) with an unknown structure field number. Mismatch of datafiles vs. source again. TODO: id replaced memsets in later source. TODO: have memsets on all exports and imports. * SOS: RC4 source should be tagged "1.27g RC4" (raduffy). Unfortunately the Linux client doesn't care a bit. Show History does work if from/to date differ by at least a day, and it shows the tag on code/ (only that subtree), but recursive get aborts halfway. Manual: http://www.sourcegear.com/SOS/Doc/ 2000-12-06 Bernd Kreimeier * TEST: accepting missing shaders now. No bots, but I can actually enter the game and play (more than can be said for classic Q3 right now). TODO: Error: BotStartFrame: bot library used before being setup * code/renderer/tr_shader.c: took out assertion for now... * TEST: now missiopack/cgame loads TODO: tr_shader.c:2275: R_FindShaderByName: failed TODO: searches ui/assets/3_cursor2.TGA, has ui/assets/3_cursor3.tga * code/unix/Makefile (MPCGOBJ): ui_shared.o (duh). DONE: /cgamei386.so: undefined symbol: PC_Float_Parse * code/botlib/be_ai_goal.c: initialize campspots etc. This might or might not fix this one (didn't get back to gdb due to mouse-only navigation). DONE: 0x80d1d5b in BotFreeInfoEntities () at be_ai_goal.c:447 * TEST: this time with missionpack/cgame loading... noy TODO: TA menu blocked after end of intro movie TODO: console in_mouse 1 doesn't grap pointer even on vid_start? * code/cgame/cg_newdraw.c: -Werror. * code/unix/Makefile (MPCGOBJ): cg_newdraw.o was missing (duh). DONE: missionpack/cgamei386.so: undefined symbol: CG_OwnerDrawVisible" * code/ui/ui_shared.c:1309 assign after bail on NULL. DONE: segfault in Item_SetFocus (item=0x0, x=0, y=0) * TEST: new set of DLL's (this time hopefully correct). All baseq3/ DLL's load, as does the missionpack/ UI DLL. The menus now work in both (TA seems mouse-only on everything but "Quit"). Segfault on delayed TA "Quit" (stack fubared): #5 0x809fc28 in VM_Call (vm=0x88408a0, callnum=3) at ..//qcommon/vm.c:617 #6 0x805aafc in CL_KeyEvent (key=9, down=qtrue, time=128644) cl_keys.c:1194 TODO: TA menu's w/o mouse? TODO: Win32 goes submenus but does not unfold TODO: Linux does not go submenus * code/ui/ui_main.c: see below. TODO: LCC gets fits - operands of = have illegal types TODO: 'pointer to const unsigned char' and 'pointer to const char' * code/ui/ui_shared.c: see below. * code/ui/ui_gameinfo.c: see below. * code/ui/ui_atoms.c: see below. * code/game/g_bot.c: more cruft. * code/cgame/cg_draw.c: loads of functions modified for MISSIONPACK that aren't used at all for MISSIONPACK anymore. Development relics. * code/cgame/cg_consolecmds.c: -Werror. Note: due to Makefile error never ever compiled... * code/unix/Makefile: fixed various dependency errors for game and ui library. TODO: create a new Makefile with patsubst and rules. TODO: why C_ONLY in the i386 dedicated server? * code/unix/unix_main.c: use dlerror() excessively. Littered more unused DLL related functions with assert(0). TODO: clean up Sys_Load/UnloadDll (a real mess) TODO: remove Zoid code cruft (unused per-DLL functions) * code/game/bg_misc.c: changed G_Printf for Com_Printf. This was undefined in baseq3/uii386.so preventing loading. * TEST: +set sv_pure 0 +set vm_game 0 +set vm_cgame 0 +set vm_ui 0 Note: so far I used only the game DLL.. duh. UI DLL fails to load: missing G_Printf. * code/unix/Makefile: -DMALLOC_CHECK in addition to the -DZONE_DEBUG I have used since switching to calloc. Using MALLOC_CHECK=1 for now, might use 2 if something comes up. * code/renderer/tr_init.c (GL_SetDefaultState): it does get called, but does not show up in the log. * TEST: tried executing a script - get bounced. TODO: is there any way to jump into a map? TODO: cl_cinematics 0 (supress all fullscreen RoQ) Next: used r_logfile 200 in Win32 (RC4) and Linux. There is a buckload of setup code seemingly not done at all in Linux? Either that, or logging is enabled with a delay in Linux. * code/unix/linux_glimp.c: fixed autorepeat (H2/Fakk2 way). 2000-12-05 Bernd Kreimeier * code/renderer/tr_mesh.c: added assert there. * TEST: menus and in-game drawing are just as they were with the initial SOS001119 port. In addition: R_AddMD3Surfaces: no such frame 0 to -2147483477 for 'models/players/xaero/upper.md3' R_AddMD3Surfaces: no such frame -2147483477 to 171 R_AddMD3Surfaces: no such frame 171 to -2147483498 ad nauseam (used as my player model). Triggered: haveing a trRefEntity_t *) 0x41dbbd00 with frame = -2147483477. Might be a red herring (PRINT_DEVELOPER), ignore for now. * code/ui/ui_main.c: missing return. * code/ui/ui_shared.c: excess byte in initializer (which gcc did not caught, but LCC did). Also LCC complains about missing returns, but gcc doesn't (neither says unreacheable code though). If necessary (MsVC?) guard with Q3_VM. * code/q3_ui/ui_ingame.c: see below. * code/q3_ui/ui_atoms.c: voidfunc_f. LCC warns about conversion from `pointer to void' to `pointer to void function(void)' being compiler dependent. Casting NULL. Guess what, doesn't fix it either. TODO: do not use these cursed scripts to generate VM code, we do not have proper rules for LCC/q3asm, thus the files never get updated. * code/unix/Makefile: for paranoia's sake recreated the 1.17 compile for the UI DLL (where only q_shared/math were actually compiled as DO_SHLIB_CC. Later: switched to different gcc. * STATIC: remaining problems are vmMain (same entry point for all DLL's), could use cgMain, uiMain and gMain here for HARD_LINKED. Note: I don't think id has used this in ages. Plus all the collisions in *_syscalls.c, which simply can't be fixed cheaply. None is the superset of 2 others, neither seems w/o overlap to others. Full stop. * code/botlib/be_aas_move.c: see below. * code/game/ai_dmq3.c: VEC_UP/DOWN, MOVEDIR_UP/DOWN now static. See also game/g_utils.c for existing static duplicates. * code/game/q_shared.h: #define stricmp strcasecmp * code/unix/Makefile: no mo' -Dstricmp=strcasecmp, see q_shared.h Also: no mo' -I/usr/include/glide, no FX TODO: are we building against system GL headers? ../Mesa/? * code/q3_ui/ui_atoms.c: comment on duplication * code/cgame/cg_drawtools.c: use UI/CGAME_HARD_LINKED on UI duplicates TODO: does this UI_ code in cg_drawtools/ui_atoms belong into ui_shared? * code/unix/Makefile: use -DQ3_STATIC * code/game/q_shared.h (*_HARD_LINKED): trigger on Q3_STATIC Later: collision between UI and CGAME is still there. This fixed the Com_Error, Com_Printf issues though * code/unix/Makefile ($(B)/q3static/ai_vcmd.o): this file was missing, hence undefined symbol. ($(B)/baseq3/game/ai_vcmd.o): same here. ($(B)/missionpack/game/ai_vcmd.o): same here. * STATIC: cg_syscalls.c, g_syscalls.c and ui_syscalls.c alias. Multiply defined symbols: Com_Error, Com_Printf VEC_UP, VEC_DOWN MOVEDIR_UP, MOVEDIR_DOWN vmMain dllEntry PASSFLOAT trap_Error trap_Milliseconds trap_Argc trap_Argv trap_FS_FOpenFile trap_FS_Read trap_FS_Write trap_FS_FCloseFile trap_FS_GetFileList trap_R_RegisterModel trap_R_RegisterSkin trap_R_RegisterFont trap_R_RegisterShaderNoMip trap_R_ClearScene trap_R_AddRefEntityToScene trap_R_AddPolyToScene trap_R_AddLightToScene trap_R_RenderScene trap_R_SetColor trap_R_DrawStretchPic trap_R_ModelBounds trap_UpdateScree trap_S_StartLocalSound trap_S_RegisterSound trap_Key_IsDown trap_Key_GetCatcher trap_Key_SetCatcher trap_GetGlconfig trap_PC_AddGlobalDefine trap_PC_LoadSource trap_PC_FreeSource trap_PC_FreeSource trap_PC_ReadToken trap_PC_SourceFileAndLine trap_S_StopBackgroundTrack trap_S_StartBackgroundTrack trap_RealTime trap_CIN_PlayCinematic trap_CIN_StopCinematic trap_CIN_RunCinematic trap_CIN_DrawCinematic trap_CIN_SetExtents trap_MemoryRemaining trap_SendConsoleCommand trap_Cvar_Register trap_Cvar_Update trap_Cvar_Set trap_Cvar_VariableValue trap_Cvar_VariableStringBuffer trap_RealTime trap_SnapVector // used in game/bg_*.c, needs conditional More aliasing between ui_atoms.c and cg_drawtools.c: UI_DrawBannerString UI_ProportionalStringWidth UI_ProportionalSizeScale Undefined symbol: ai_team.o: In function `FindHumanTeamLeader': ai_team.c:1899: undefined reference to `BotVoiceChat_Defend' Note: * code/game/g_main.c: unused. * code/game/g_arenas.c: unused. * code/game/ai_team.c: init. * code/game/ai_dmnet.c: /* in comment (odd). Note: why do these come up now but not earlier? TODO: the make dependencies might target wrong files. * code/unix/Makefile (TARGETS): added q3static. Note: this is baseq3/ * TEST: +set r_logfile 100. It seems that the addition of code (add an assertion etc.) changes the behavio of the binary. The intro cinematics code seems to suffer first - didn't play, then played, then (another assert added) doesn't play. Watch out for (missionpack): UI_CIN_PlayCinematic SCR_PlayCinematic( mpintro.roq ) trFMV::play(), playing mpintro.roq Also fails to exit cleanly: break gives #0 0x401919ee in __select () #1 0x400bbcb8 in __DTOR_END__ () #2 0x4004baa1 in _XSend () #3 0x452b009f in GLXRenderFlush () #4 0x804ce0c in _XRead () #5 0x40680813 in ?? () Stack is corrupted. Note: ~/.q3a/gl.log TODO: write per-frame files (see Heretic2) TODO: add Heretic2 QGL (more detail) * code/unix/linux_qgl.c (QGL_EnableLogging): fixed countdown (i.e. propagated changes from win32/, see Fakk2). * code/unix/linux_glimp.c: fixed QGL_EnableLogging argument to avoid cast error (always qfalse). * code/unix/Makefile (DEBUG_CFLAGS): use ZONE_DEBUG. * code/qcommon/common.c: replaced malloc with calloc calls. * code/q3_ui/ui_local.h: have to use ui/ui_public.h * code/cgame/cg_servercmds.c: requires ../ui/menudef.h * code/cgame/cg_consolecmds.c: ui/ui_shared.h is unique. * code/q3_ui/ui_public.h: make sure this won't be compiled. * code/client/client.h: we have to include ui/ui_public.h. Note: id is obviously maintaing only the ui/ headers, so the headers in q3_ui/ are deprecated. * code/renderer/tr_shader.c: added assertions (see Ryan's Fakk2 problems with missing shaders). * code/game/g_cmds.c: below. * code/game/ai_vcmd.c: below. * code/game/ai_team.c: below. * code/game/ai_dmnet.c: below. * code/game/ai_dmq3.c: below. * code/game/ai_chat.c: below. * code/game/ai_cmd.c: ../../ui/menudef.h (new Q3TA script directory). * code/cgame/cg_newdraw.c: make sure it won't compile w/o MISSIONPACK. * code/cgame/cg_servercmds.c: below. * code/cgame/cg_event.c: below. * code/cgame/cg_consolecmds.c: below. * code/client/keys.h: below. * code/client/client.h: below. * code/q3_ui/ui_local.h: include from ../q3_ui/ not ../ui/. Note: id seems to intentionally use the header from the new ui/. * Makefile: checked -I$(UIDIR), there is no such. That means all files include directly, which means all (including Q3) are using the new ui/ headers. 2000-12-04 Bernd Kreimeier * RC4: released as 362101115 Dec 4 11:40 TA_Q3A_RC4.zip * TEST: the corrupted menu problem is back :-(. Looks like I am in for a static link next. * code/unix/Makefile (clean2): fixed (not all new OBJ covered). * code/q3_ui/ui_teamorders.c: -Werror. * code/q3_ui/ui_team.c: -Werror. * code/q3_ui/ui_qmenu.c (Bitmap_Draw): -Werror. * code/q3_ui/ui_mods.c (UI_Mods_LoadModsFromFile): unused. -Werror. * code/q3_ui/ui_controls2.c: -Werror. * code/q3_ui/ui_atoms.c: -Werror * code/null/null_client.c: -Werror. * code/unix/linux_joystick.c: -Werror. * code/unix/linux_glimp.c: -Werror. * code/unix/linux_qgl.c: -Werror. * code/unix/unix_shared.c: -Werror. * code/unix/unix_net.c: -Werror. * code/unix/linux_local.h: added missing prototypes. * code/unix/unix_main.c: -Werror. Includes linux_local.h * code/jpeg-6/jdmainct.c: see below. * code/jpeg-6/jcmainct.c: variables called "main" (*moan*) * code/jpeg-6/jcdctmgr.c (forward_DCT): -Werror. * code/botlib/l_script.c (PS_ReadLiteral): -Werror * code/botlib/l_precomp.c (PC_AddBuiltinDefines): -Werror. * code/botlib/be_interface.c: -Werror. * code/botlib/be_aas_reach.c: -Werror * code/botlib/be_aas_cluster.c: -Werror * code/game/be_aas.h: -Werror. Note: MrElusive accumulates a lot of code history in nested comments, which gcc doesn't like at all. #if 0'ed to avoid. * code/qcommon/vm_interpreted.c: -Werror. * code/qcommon/unzip.c: -Werror. * code/cgame/cg_servercmds.c: -Werror. * code/cgame/cg_main.c: -Werror. * code/cgame/cg_drawtools.c: -Werror. * code/game/bg_misc.c: -Werror. * code/game/be_ai_move.h (bot_avoidspot_s): added. * code/botlib/be_ai_move.c: removed typedef struct bot_avoidspot_s * code/client/snd_mix.c: -Werror. * code/qcommon/md4.c: -Werror. * code/qcommon/common.c: -Werror. * code/client/cl_keys.c: -Werror. * code/client/cl_cin.c: -Werror, init local variables. * code/unix/Makefile: -Werror. need -O for -Wall for uninitialized Note: the above is the list of files that got touched during a pass with -g -O -Werror -Wall flags (in the hope of finding uninitialized memory and ambiguous statements). Most of the above are simply unused variables (or even code). TEST: RC3 data files, but DLL's. TODO: TA gets stuck in initial sound, doesn't play cinematics (sometimes) TODO: Q3 intro movie looses sound after Sarge gets teleported TODO: Q3 ingame renders world, weapon, muzzleflash, hud, can shoot, TODO: but no movement, hud background is fubared. * code/cgame/cg_main.c: cg_singlePlayerActive * code/q3_ui/ui_login.c: doesn't seem to be used? * code/game/g_rankings.c (G_RankRunFrame): doesn't seem to be used. * code/q3_ui/ui.sh: disabled this. * code/q3_ui/q3_ui.sh: changed include path to ../q3_ui/ (duh). * code/game/game.sh: changed include path to ../q3_ui/ which is not in the Win32 batch file. * code/cgame/cg_rankings.c: this does not seem to be included. * code/cgame/cgame_ta.sh: added -DCGAME. Also added cg_syscalls.c to build (also missing in Win32). * code/cgame/cgame.sh: added -DCGAME (see cgame.bat). Also changed include path to ../q3_ui/ which is not in the Win32 batch file. Also added cg_syscalls.c to build (missing in Win32). 2000-12-01 Bernd Kreimeier * RC3: released as of sos001201 / Q3 1.27f * code/unix/Makefile: more fixes with clean build. The changes made fix the menu rendering for Q3 but not TA. Ingame graphics still broken. * code/game/game_ta.sh: created. Use game_ta.q3asm here. * code/game/game.sh: no -DMISSIONPACK * code/game/game_ta.q3asm: CR/LF, /. * code/cgame/cgame_ta.sh: created. Use cgame_ta.q3asm here. * code/cgame/cgame.sh: no -DMISSIONPACK. No cg_newdraw, ui_shared. * code/cgame/cgame.q3asm: No cg_newdraw, ui_shared. CR/LF, /, cg_newDraw, and the output path/name. * code/q3_ui/q3_ui.q3asm: output to ui not q3_ui... * code/cgame/cg_event.c: cg_singlePlayerActive used here. TODO: guard by MISSIONPACK * code/cgame/cg_local.h: named q3print_t enum. Cvar cg_singlePlayerActive for both Q3 and TA. * code/unix/Makefile: cleanly separate B/baseq3/ and B/missionpack/ subtrees during build. While new and old UI are in separate directories, the cgame/ and game/ are shared, with conditional -DMISSIONPACK compile and different files includeds (cd_draw, cg_newdraw). That means twice the number of targets (3 DLL's, 3 QVM's, times two), and different build rules. TODO: carefully check Win32 build for (other) conditionals TODO: carefully check Win32 build for link lists * CVS: ui/, code/ui, botfiles/ and subdirectories are added. The code/macosx/ directory turned out to be a real pain that had to be edited manually, throwing out CVS/ directories in the tree that had been created by SOS as they are in id's repository: code/macosx/Client/CVS code/macosx/Client/PBUserInfo/CVS code/macosx/Client/Quake3.nib/CVS code/macosx/Common/CVS code/macosx/DedicatedServer/CVS code/macosx/DedicatedServer/PBUserInfo/CVS Now tagged cvs tag id1-26y-loki001119 TODO: there are several new files not yet linked? * ChangeLog: merged the Changelog from the bk00119 working branch (initial Q3TA port) based on sos001119 snapshot. Also merged the source tree with cvs-1.17. In the ChangeLog below *** MISSIONPACK *** indicates work that was done on the branch (code-sos/ prefix in files). The cvs update of this will be tagged with cvs tag id1-26y-loki001119 Use this tag to hunt for possible Linux fixes that got lost (i.e. got dropped by id since id000516 and were thus not in sos001119, but did not show in diff id000516 cvs1-17). New directories in CVS: botfiles/, ui/. Missing from SOS/Missionpack: SDK directories. common, lcc, libs, q3asm, q3data, q3map, q3radiant. * ssreport.txt: below. Note: watch for files called "ssreport.txt", that's id ChangeLog. * ui/ui_syscalls.asm: below. * q3_ui/ui_syscalls.asm: below. * game/g_syscalls.asm: below. * cgame/cg_syscalls.asm: below. * bspc/linux-i386.mak: below. * bspc/lcc.mak: below. * botlib/linux-i386.mak: below. * botlib/lcc.mak: below. * A3D/a3d_console_variables.txt: CR/LF issue (minimize diffs). * CVS: the checked bk001119 work copy of the sos001119 initial checkout (completed with everything in the SOS "Missionpack" tree, i.e. botfiles/ and botfiles.* added), copied over the cvs-1.17 checkout. Note: in these cases, BEWARE ui -> q3_ui/ links, and different ChangeLogs. Also "make clean" helps. * unix/unix_net.c: below. * unix/unix_main.c: below. * unix/matha.s: below. * unix/linux_qgl.c: below. * unix/linux_glimp.c: see also linux_joystick.c. * server/sv_client.c: below. * renderer/tr_surface.c: below. * renderer/qgl.h: below. * qcommon/qcommon.h: below. * qcommon/files.c: below. * qcommon/common.c: below. * q3_ui/ui_demo2.c: below. * mac/mac_net.c: below. * mac/mac_glimp2.c: below. * game/surfaceflags.h: below. * game/bg_lib.c: checked against id00516/cvs-1.17a diff. * bspc/bspc.c: TH_AASToTetrahedrons call removed since id000516. Note: our final compare of id000516 against cvs-1.17a, making sure that all these differences are in bk001119 (initial Q3TA port). If id branched the Q3TA base off before id000516 we might be screwed. Note: I do not diff against bk000520, which had some minor changes against id000516 (check VectorArrayNormalize, OTConfiguration), which seem consistent with me taking a pre-id000516 source snapshot for that working branch. 2000-11-30 Bernd Kreimeier * TEST: compiled using the symbolic link ui/ -> q3_ui/. Had to undo one CVS change, regarding code/cgame/cg_syscalls.asm code/game/g_syscalls.asm code/q3_ui/ui_syscalls.asm These files are neither generated by Win32 cgame.bat nor cgame.sh (etc.), thus seemingly maintained by hand. cvs tag pr1-17-loki001130b should be used if somebody needs this 1.17 snapshot (which, remember, is post-release, with additional fixes). Later: cvs tag pr1-17-loki001130c includes the full ChangeLog (duh). * CVS: up until cvs-1.17-001130, code/ui/ contained the Q3 code for the UI QVM/DLL. In Q3TA, this code has been moved to code/q3_ui/, while at the same time the new (scripting driven) UI code for Q3TA was maintained in code/ui/. To preserve the history of code/ui/, it has been renamed to q3/ui/ in the CVSROOT. Note: this will BREAK all cvs-1.17 and before checkouts. To compile earlier versions, move or link q3_ui/ to ui/. The code has been tagged cvs tag pr1-17-loki001130 after the change. DONE: remove code/*/vm/*.asm from CVSROOT Note: this includes code/*/*.asm files (from *_syscalls.c). These were originally tracked in CVS, but if we need comparison of q3asm output or QVM files we can rely on the Win32 and Linux SDK now. These files have been physically removed from CVS now, followed by cvs tag pr1-17-loki001130a 2000-11-30 Bernd Kreimeier *** MISSIONPACK *** * RC2: new ZIP file (another 360M for convenience). * SOS: new CVS module, Quake3_sos. This will be used to track the unchanged SOS checkouts from id. As their repository is read-only, and there is no estimate on when changes might be backpropagated there, I will track their changes in a separate module, and update our local Quake3 module accordingly. This is effectively "tracking 3rd party" w/o import and half-automated, forced mergers - in other words, we now branch starting with our post-1.17 changes, for the benefit of moving at all. Baseline is a slightly changed PR-1.17 id000516 source dump (essentially ui/ moved to q3_ui for continuity, and CR/LF etc.). cvs import Quake3_sos id000516 pr1-17-win32 cvs import Quake3_sos sos001119 pr1-26-win32 cvs import Quake3_sos sos001120 pr1-26-win32-001120 cvs import Quake3_sos sos001121 pr1-26-win32-001121 cvs import Quake3_sos sos001122 pr1-26-win32-001122 cvs import Quake3_sos sos001122-demo pr1-26-win32-demo This is about the 1.26w Team Arena Win32 demo release, give or take a couple of lines. Has Q3_VERSION "Q3 Team Arena Demo 1.27b". cvs import Quake3_sos sos001123 pr1-26-win32-001123 cvs import Quake3_sos sos001126 pr1-26-win32-001126 Now track id versions (see code/game/q_shared.h:Q3_VERSION) cvs import Quake3_sos sos001128 pr1-27c-win32-001128 With 1.27d they switched from Demo to full version (RC1). cvs import Quake3_sos sos001129 pr1-27d-win32-001129 cvs import Quake3_sos sos001130a pr1-27d-win32-001130a Now switched to 1.27e. This import is done from the SOS working directory. cvs import Quake3_sos sos001130b pr1-27e-win32-001130b Note: SoS created rwx attributes which are luckily fixed automagically during import. It is also seemingly incapable to compare files, and leave files that have not changed the hell alone. I can't do cvs update due to the $..$ tags in the original files (which CVS can't be told to ignore), so I have to do import (creating a load of vendor tagged branches), but at least cvsweb and cvs get the revisions right. * code-sos/unix/Makefile: added linux_joystick * code-sos/unix/linux_local.h: match mac/ and win32/, for prototypes. * code-sos/unix/linux_joystick.c: new file, code from linux_glimp.c Note: decided to separate this, as (a) we might edit/extend a lot, (b), it's not in the id tree, (c) it's not GL, (d) there might be even more oddball devices. Anything that cuts down on diffs. * code-sos/unix/linux_glimp.c (Q_stristr): const return (cvs1.17). Also (XLateKey): added more keyboard mappings (ASCII on upper row digits) (cvs1.17). Added in the minimal joystick hooks (cvars, function calls). Fixed joystick cvar naming to match win32 (kept joystick_threshold). TODO: joystick stubs for dedicated? * CVS: I have to move up to 1.27d (data, Win32 networking). With exception of linux_glimp.c (mostly joystick code), all cvs1.17 changes should now be in the work snapshot based on the first sos001119 we got from id. There are also some additional changes in there already, thus I'll move the (buggy) 1.26 snapshot into CVS before adding even more differences. 2000-11-29 Bernd Kreimeier *** MISSIONPACK *** * RC1: TeamArena_Q3A_RC1.zip. Source has moved from Q3VERSION "Q3 Team Arena Demo 1.27c" to "Q3 1.27d" now. * code-sos/qcommon/common.c: added Com_InitPushEvent(). Also increased MAX_PUSHED_EVENTS to 256. Note: this is another case of buffer memory not zero'ed. Com_EventLoop, fixed evTime to evType in debug print. * TEST: baseq3/ +set sv_pure 0 +set vm_game 0 +set in_mouse 0 +set developer 2 TODO: Team Arena in menu leads to RE_Shutdown(1) and locks TODO: can't play game TODO: shaders can't load *.tga, *.jpg files are there TODO: DO_CC linking for DLL's, DO_SHLIB_CC only for export? TODO: ERROR: Bad player movement angle TODO: Warning: cvar "..." given initial values: "..." and "..." TODO: TA demo ERROR: CL_ParseServerMessage: Illegible server message TODO: WARNING: Com_PushEvent overflow * code-sos/qcommon/files.c: add NULL filter for our Sys_ListFiles calls. * unix/unix_shared.c (Sys_ListFiles): signature has changed, additional Sys_ListFiles argument now. * code-sos/unix/unix_net.c (Sys_GetPacket): see below (readcount=0). * code-sos/unix/unix_main.c: see below (Mike's and my changes to DLL loading, my event buffer clear fixes). * code-sos/unix/linux_qgl.c (QGL_Init): see below (__FX__ guards). TODO: abstract WGL/GLX and end unfortunate QGL duplication. TODO: spice up QGL with Linux H2 full version. * code-sos/q3_ui/ui_demo2.c: fix on demo names - no Q_strupr(demoname). Note: in CVS this fix is in ui/ui_demo2.c. CVS is screwed by id choosing the old name for new directory... TODO: manual intervention on "ui goes q3_ui" in CVSROOT? * renderer/qgl.h: see below (__FX__ guards). * qcommon/files.c: migrated in the 1.17cvs changes against the id000516 code dump, i.e. the (not marked - boo hiss) mkv changes. Note: all the above is based on a diff of the last id code dump pre-1.17 against our CVS, with those fixes now migrated into the sos1.26 snapshot. TODO: move in joystick code. TODO: replace XAutoRepeatOn/Off with filter (focus). TODO: DGA 2.0 and such. * code-sos/game/q_shared.c: valid compare for NULL strings * code-sos/unix/unix_main.c: QRTLD, and now using RTLD_NOW. Note: it is a bad idea to load game DLL's that are missing symbols. * code-sos/ui/ui_main.c: see below. * code-sos/game/g_main.c: see below. * code-sos/q3_ui/ui_main.c: see below. * code-sos/cgame/cg_main.c: made cvarTable and cvarTableSize static. This resolved a segfault related to traversing the UI table during Init. Note: there is a segfault related to this variable being out of bounds. Different struct size in global variables possible aliasing between the DLL's. * code-sos/unix/unix_main.c (Sys_Error): assert(0), no exit in debug. * code-sos/game/q_shared.c: now aborts on NULL destination. Also DPrintf's on bogus excess copies. TODO: make all those string functions safe, at least assert. * code-sos/server/sv_init.c: comment in SV_Init // init the botlib here because we need the pre-compiler in the UI Called in qcommon/common.c:Com_Init, were CL_Init is called afterwards... * code-sos/server/sv_bot.c: the botlib_import is filled here. * code-sos/unix/unix_main.c (Sys_GetBotLibAPI): RTLD_NOW. Which is for naught, as this code is not used and has never been used. assert(0) * code-sos/botlib/be_interface.c: botimport supposed to be set here. * code-sos/botlib/l_memory.c: segfault with q3_ui/ DLL. #1 0x80e23ec in GetMemory (size=35) at ..//botlib/l_memory.c:331 331 ptr = botimport.GetMemory(size + sizeof(unsigned long int)); as botimport is completely NULL'ed. * code-sos/q3_ui/q3_ui.sh: created from ui/ui.sh 1.17 * code-sos/q3_ui/q3_ui.q3asm: unfubared (CR/LF, / path). * code-sos/unix/Makefile: added q3_ui/ make targets (basically ui/ targets from CVS 1.17 Makefile for starters). * code-sos/q3_ui/: this is the old UI code, which does not use ../ui/menus.txt (see ui/ui_main.c). In other words, the code in ui/ now has to be compiled or qvm'ed for missionpack/, but to create the necessary DLL or QVM modules for baseq3/ we need to use q3_ui/. 2000-11-27 Bernd Kreimeier *** MISSIONPACK *** * code-sos/game/bg_lib.c: ld problem with a custom "tan(..)" TODO: loooking forward to SIGFPE on this code base. * code-sos/ui/ui_util.c: this file is empty. * code-sos/ui/ui.sh: new files: ui_shared.c ui_util.c Replaced by the /ui/*.menu files: q3lcc: can't find `../ui_cdkey.c' q3lcc: can't find `../ui_ingame.c' etc. * code-sos/cgame/cgame.q3asm: added cg_newdraw entry. Also added ui_shared entry. * cgame/cgame.sh: added cg_newdraw.c entry. Also added ../ui/ui_shared.c entry. * code-sos/cgame/cg_newdraw.c: renamed (was cg_newDraw.c mixed case). Note: the infidels have taken over. * cgame/cgame.sh: added -DMISSIONPACK. Note: w/o, q3lcc complains ../cg_event.c:204: undeclared identifier `cg_singlePlayerActive' ../cg_event.c:204: left operand of . has incompatible type `int' which indicates that this source does not compile w/o MISSIONPACK anymore. The baseq3/pak4.pk3 file in the Q3TA snapshot archives are dated 284464 11-10-00 14:02 vm/cgame.qvm 463940 11-14-00 14:47 vm/qagame.qvm 271596 11-14-00 14:48 vm/ui.qvm the code dump is from 11-19. Note: Make does not abort on q3lcc complains * code-sos/game/game.sh: also added ai_vcmd.c entry. * code-sos/ui/ui.q3asm: fubared (below). In addition, this is the only one to have a -o "/tmp/quake3/missionpack/vm/ui" line in it. Given that the other 2 QVM modules are also dependend on -DMISSIONPACK, this seems a real mess. For now using the same path as the other 3. * code-sos/cgame/cgame.q3asm: below. * code-sos/game/game.q3asm: fubared. Fixed CR/LF and \ in paths again (read by q3asm called by game.sh called by make). * unix/Makefile: updated fpor DLL/QVM. Note: also shell scripts to use q3lcc not lcc. 2000-11-27 Bernd Kreimeier * code/unix/Makefile: now expects a run/ directory relative (between this, the Loki standards, and the utility code in the same repository, it's ever so slightly less dorky). TODO: fix broken copyfiles target etc.pp. * code/game/bg_lib.c: turns out the changes I undid 001120 were affecting original Zoid Linux port related defines, which break VM compile. Mike fixed those (which I unfixed when referring to the latest id code that does not contain these patches). However, they duplicate ANSI libc symbols, so the guards might be wrong. The symbols are missing when compiling for VM, so I now use the existing lcc -DQ3_VM flag: //#if !defined ( _MSC_VER ) && ! defined ( __linux__ ) #if defined ( Q3_VM ) This will break DLL compile on non-ANSI platforms, which will have to be added to the conditional then. * code/ui/ui.sh: below. * code/game/game.sh: below. * code/cgame/cgame.sh: Linux SDK installs q3lcc to avoid collisions with regular lcc pre-installs. The scripts fail with "lcc not found", but do not abort the Makefile. Note: now that VM code gets actually built, there are errors: g_main:648 ERROR: symbol vsprintf undefined bg_pmove:1221 ERROR: symbol abs undefined q_math:4309 ERROR: symbol fabs undefined q_shared:2801 ERROR: symbol tolower undefined q_shared:2862 ERROR: symbol toupper undefined ai_dmq3:208 ERROR: symbol atoi undefined ai_cmd:4951 ERROR: symbol sscanf undefined 2000-11-20 Bernd Kreimeier * TEST: test compile of pr-1.17+cvs fixes segfaults due to new baseq3/pak4.pk3 Note: to self ... 1.17 is not compatible with new files. Checking into CVS next. * code/: changes applied by us that are not in id's code base affect q_shared.c (NULL in Q_stricmp), files.c (FIXME fs_cdpath, Sys_ConcatenateFileList, ui_demo2.c (demo no tolower on linux). In unix/ linux_glimp.c (joystick code), qgl.h, linux_qgl.c (__FX__), unix_main.c (dlopen bug and event buffers), unix_net.c (readcount), matha.s (assembly warning). * code/server/sv_client.c (SV_WriteDownloadToClient): No effective change on FS_SV_FOpenFileRead call, they reworked autodownload some more seemingly. * code/renderer/tr_surface.c: VectorArrayNormalize * code/qcommon/qcommon.h: see below. * code/qcommon/files.c: Com_ReadConfigs removed. * code/qcommon/common.c: removed Com_ReadConfigs, textual replacement of body in Com_Init. * code/mac/mac_net.c: not applied (undone by id) OTConfiguration *config <> OTConfigurationRef config * code/mac/mac_glimp2.c: r_colorbits->integer > 16 * code/game/surfaceflags.h (CONTENTS_BOTCLIP): added. * code/game/q_shared.h: not applied (undone by id) #if defined(ppc) || defined(__ppc) || defined(__ppc__) #define idppc 1 #else #define idppc 0 #endif * code/game/q_math.c: added another CPP line to guard BoxOnPlaneSide, removed WIN32 guard. TODO: this could be broken code guarded in all current compiles... * code/game/bg_lib.c: left Q#_VM guard for typedef cmp_t Added !defined( __linux__ ) for tolower and atoi. Note: the changes above relate to the very last code update from id prior to the 6 month blackout, which were not in CVS when Michael made his updates. Needed to establish the baseline for the new patch. Source dump 1.17.00520, against SOS 1.26w-001119 version. 2000-11-20 Bernd Kreimeier *** MISSIONPACK *** * TEST: running against the data up to TeamArena_Q3A_001109.zip Hunk_Clear: reset the hunk ok Program received signal SIGBUS, Bus error. "q3dm2", killBots==qtrue #0 CM_ClearMap () at ..//qcommon/cm_load.c:644 #1 0x80884a7 in SV_Map_f () at ..//server/sv_ccmds.c:159 #2 0x8072579 in Cmd_ExecuteString (text=0xbffff4b0 "spmap q3dm2") at ..//qcommon/cmd.c:591 #3 0x8071dfe in Cbuf_Execute () at ..//qcommon/cmd.c:190 #4 0x80763f7 in Com_Frame () at ..//qcommon/common.c:2547 #5 0x8130d6b in main (argc=13, argv=0xbffff984) at ..//unix/unix_main.c:953 #6 0x40100cb3 in __libc_start_main (main=0x8130bc4
Not reproducible (screen stayed black). * TEST: +set developer 1, same for Win32 and Linux: Can't find gfx/misc/flare.tga Can't find gfx/misc/sun.tga Can't find gfx/misc/console02.tga Can't find vm/ui.map Can't find textures/sfx/logo512.tga Can't find gfx/colors/black.tga Can't find models/mapobjects/banner/banner5_2.md3 Can't find models/mapobjects/banner/banner5_1.md3 Can't find textures/sfx/firegorre2.tga Can't find textures/sfx/bolts.tga Can't find menu/art/unknownmap.tga * Q3TA: after nearly 6 months, a code update from id. SOS access even. Got it to compile, link and start, but its currently broken (menu doesn't render in full, can't get into game etc.). Need a baseline 1.17 to diff against. Last code dump was May 16, with bspc code updated May 19. Checking working directory of bk000520 against CVS next (Mike's fixes never made it into id's codebase or a post 1.17 release, neither did my fixes as released in the point release version 1.17). 2000-11-19 Bernd Kreimeier *** MISSIONPACK *** * TEST: Win32 install as tested with 1.26w. quake3.x86 (Q3A game) Warning: cvar "r_uifullscreen" given initial values: "1" and "0" Warning: cvar "r_inGameVideo" given initial values: "1" and "0" ^3WARNING: sound/feedback/hit.wav is a 8 bit wav file (on windows, sound/weapons/weapon_hover.wav is missing...) Menu only partially displayed in TA and baseq3 play, menu itself seems to work. Freetype? WARNING: Com_PushEvent overflow * code-sos/game/game.sh: not in SOS, moved in from CVS snapshot. * code-sos/qcommon/common.c: conditional DEDICATED to get rid off CL_ShutdownCGame/CL_ShutdownUI/CIN_CloseAllVideos. Same for UI_usesUniqueCDKey: dedicated server does not write CD key file. TODO: check whether there is an unneeded "read CD key" for dedicated server. * code-sos/null/null_client.c (CL_ShutdownAll): added dummy. * code-sos/unix/Makefile: server/sv_net_chan.o for dedicated server. * code-sos/null/null_snddma.c: fixed S_RegisterSound signature. * code-sos/client/snd_mix.c: snd_p, snd_linear_count, snd_out can't be static, as used by unix/snd_mixa.s. * code-sos/unix/Makefile: added to the executable target: renderer/tr_font.c client/cl_net_chan.c server/sv_net_chan.c Also added a lot of jc*.c files to build, to fix unresolved symbol errors. TODO: is there unused jpeg-6/jd*.o code linked in now? * code-sos/ft2/smooth.c: includes ftgrays.c, ftsmooth.c * code-sos/ft2/truetype.c: ttdriver.c, ttpload.c, ttgload.c, ttobjs.c. Also (see ftoption.h) TT_CONFIG_OPTION_BYTECODE_INTERPRETER ttinterp.c * code-sos/ft2/sfnt.c: includes ttload.c, ttcmap.c, sfobjs.c, sfdriver.c. lso (see ftoption.h) TT_CONFIG_OPTION_EMBEDDED_BITMAPS ttsbit.c TT_CONFIG_OPTION_POSTSCRIPT_NAMES ttpost.c * code-sos/ft2/ftbase.c: includes ftcalc.c, ftobjs.c, ftstream.c, ftlist.c, ftoutln.c, ftextend.c, ftnames.c. * code-sos/ft2/autohint.c: includes ahangles.c, ahglyph.c, ahglobal.c, ahhint.c, ahmodule.c. * code-sos/unix/Makefile: added ft2/ to client objects, took out ftraster.c/ftrend1.c (see below), added -DFT_FLAT_COMPILE. * ft2/ftsmooth.c: -DFT_FLAT_COMPILE required. * ft2/raster1.c: -DFT_FLAT_COMPILE required. Note: this includes ftraster.c/ftrend1.c. * code-sos/qcommon/vm_x86.c: _ftol is missing, ftolPtr only defined for Win32, but used in generic code. Workaround for now. TODO: find good Linux ftol, or use old solution. * SoS checkout. chown -R a+w * recode ibmpc:lat1 */*.h */*.c 2000-06-30 Michael Vance * misc: Spoke with Leonardo about qvm mess. * ui/ui.sh: Created to build much like the ui.bat script. * ui/ui.q3asm: Use linux style paths. * game/game.sh: Created to build much like the game.bat script. * game/game.q3asm: Use linux style paths. * cgame/cgame.sh: Created to build much like the cgame.bat script. * cgame/cgame.q3asm: Use linux systel paths. * unix/Makefile: Use the new .sh scripts to build the QVM files. * lcc/etc/linux.c: Build .asm files instead of .s files. * misc: QVMs now load properly, with minor glitches that should hopefully be solvable. The new build scripts conflict with the .asm files already in CVS, as the generated byte code is slightly different in some cases. 2000-06-29 Michael Vance * lcc/makefile: Tweaked to automatically include the system compiler's header location. Added an install directory. * lcc/custom.mk: Added a build directory. * lcc/etc/linux.c: Numerous small tweaks to make compiling the VM code a much simpler task. * q3asm/Makefile: Created. * q3asm/q3asm.c: Fixed uninitialized variable in HashString(). Fixed off by one in argument parsing. * misc: Had Brian remove the Xmd.h include from glx.h so that we can build Quake3 on XFree86 4.0 systems. * wine: Attempted to build with lcc.exe and q3asm.exe using wine, also did not work. This is in contrast to MikeP's .qvms, which seem to work. 2000-06-28 Michael Vance * common/files.c: Fixed Mods menu behaviour. * unix/linux_qgl.c: Guarded references to fxMesa. * renderer/qgl.h: Guarded references to fxMesa. * ui/ui_demo2.c: Don't convert filename to uppercase. 2000-05-07 Bernd Kreimeier * common/cmdlib.c: windowism, not guarded. Added WIN32 around "ATOM a". * q3map/Makefile: Linux Makefile. * q3map/Makefile.irix: "makefile" in original code, Irix-only Makefile. Just fixed some redundant TAB that GNU make despises about as much as I despise GNU Make, and changed to a relative path. 2000-05-01 Bernd Kreimeier * q3radiant/: updated with Q3Radiant198b3-src.zip. Tagged (globally) as q3radiant-198b3. Kept the old files 3DFXCamWnd.h 3DFXCamWnd.cpp MainFrm2.cpp New files Shaders.h misc/ (contributed special TGA resources, don't relly belong) Removed: pName Changed filenames to previous case: UNNAMED.MAP -> unnamed.map RES/BMP0002.BMP -> RES/bmp00002.bmp Changed: changelog.txt -> ChangeLog 2000-04-28 Bernd Kreimeier * CVS: bk000425 modified sources. This replaces the unix/ directory which is not yet in id's SourceSafe. Two check-ins, due to minor changes in an attempt to nail the Voodoo3 related crashes (driver problems, not a Q3 issue). Undid some of the QFL changes for PI and the log bug fix - put back in (TODO). Also includes: * Quake3/code/botlib/be_aas_sample.c: single file update from Robert. * CVS: id000423 code dumps (two of them). Applying Loki patches. Tagged for the final version (all patches). * CVS: id000422 code dump. This did not include the 1.16n fixes used for Linux, and was the first dump for the 1.17 security fix release. Note: forgot to check in the ft2/ headers themselves, but they are not used in the current codebase anyway. Are added in next dump. Also there is use of CVS/CVS-like $Keyword$ patterns in some files, and between their revisions and ours we fuck this up. Also, id ZIP files create write protected sources, have to do chmod -R a+w Quake3/ to work and overwrite files. * CVS: bk000315 modified source. This version was the 1.16n release. Note: the changes applied here are not in the subsequent code dumps of id. If you want to compile the Linux version as released you have to use bk-tagged versions until the patches are merged in by Robert Duffy. * CVS: id000314 engine code dump, same procedure as below, tag. Note: this version added vm/ sudirectories with assembly files for cgame, game, ui. CVS tag id000314. * CVS: id000304 engine code dump. Now there is a problem, as CVS was used in the Mac sources. Do find . -name 'CVS' -exec rm -r {} \; before cvs update, then tagged: cvs -d /loki/cvsroot/ tag id000304 Quake3/ * CVS: checked in a source snapshot of the id00303 engine code and the id0003029 tools code. The tool sources are not fully in sync, and we have only partial source from earlier engine revisions. The engine source marks where Loki took over from Dave Kirsch. This snapshot (with all temporary and bogus files) is imported and tagged using: cvs -d /loki/cvsroot import Quake3 id000303 initial Modules: code: the Q3 engine code, including a jpeg-6/ copy common: code shared by tools libs: code shared by tools, inlcuding a jpeg6/ copy q3asm: VM bytecode assembly q3data: misc. Q3 data conversions q3map: BSP builder q3radiant: Win32 editor, as is lcc: C compiler for q3asm The sources have not been cleaned up, and binary files have not been removed. The Q3Radiant code base might exhibit mixed case asmbiguities in the future, and future source dumps might come from SourceForge instead. 2000-04-25 Bernd Kreimeier * q3code.id000425/unix/Makefile: relative path, relocatable. Note: first code merge with id, finally :-). 2000-04-24 Bernd Kreimeier * q3code.bk000422/unix/matha.s: in C(BoxOnPlaneSide) the following line triggers assembler warning: "missing prefix `*' in absolute indirect address, maybe misassembled!" jmp Ljmptab(,%eax,4) * q3code.bk000422/unix/Makefile (MOUNT_DIR): rember to change. TODO: fix this bloody Makefile to be relocatable, damnit. * q3code.bk000422/cgame/cg_event.c: applied JCash fix again (see EV_EVENT_BITS below). Send e-mail to verify. * q3code.bk000422/renderer/tr_image.c: "../jpeg-6/jpeglib.h" again. * q3code.bk000422/: created from the id dump of today, lacking all but one of my changes (sigh). Swapped unix/ competely, takes care of 90%. Submitted all changes again to Robert... 2000-04-19 Bernd Kreimeier * q3code.bk000315/unix/linux_glimp.c (GLimp_EndFrame): QGL_EnableLogging( r_logFile->value ) doesn't work? * q3code.bk000315/unix/linux_qgl.c: GLimp_LogNewFrame() is obsolete. QGL_EnableLogging was out of sync with Win32 and did not support the new framecounter decrement logic. 2000-04-03 Bernd Kreimeier * q3code.bk000315/server/sv_snapshot.c: svs.nextSnapshotEntities is a signed integer unconditionally incremented, which gets negative and causes a segfaulting indexing an array. Added reset to counter. Might fail if snapshot numbers are supposed to monotonically increase. 2000-04-02 Bernd Kreimeier * q3code.bk000315/client/cl_parse.c (CL_ParseServerMessage): assert(0) on Illegible message (remember to +set in_mouse 0). TODO: have to add a dump message function, it's unreadable. * botlib/be_ai_goal.c (InitLevelItemHeap): loop counter -2 left -2 with uninitialized next, and -1 disconnected. Removed redundant memset. There is an item alloc leak I suspect, as max_levelitems 1024 merely delayed the overflow error. 2000-04-01 Bernd Kreimeier * botlib/be_ai_goal.c (InitLevelItemHeap): still segfaults. Not memsetting the entire item heap. As items are cleared on return, that leaves only memory corruption? Later: upped max_levelitems from 256 to 1024 Later: client dies on connect: Error: CL_ParseServerMessage: Illegible server message 255 2000-03-31 Bernd Kreimeier * botlib/be_ai_goal.c: initializing global vars. Segfault in AllocLevelItem () at /home/bk/Games/Quake3/q3code/botlib/be_ai_goal.c:364 I suspect that the initial freelevelitems setting is at the end of the list and eventually exposed. * cgame/cg_event.c: according to Johmn Cash: itemNum = (es->event & ~EV_EVENT_BITS) - EV_USE_ITEM0 Quote: "This causes itemNum to be invalid about half the time, preventing any client side effect tied to the item from occurring." 2000-03-06 Bernd Kreimeier * qcommon/common.c: set pushEvent buffer and indices to zero in Com_Init(). * q3code/qcommon/qcommon.h: made SE_NONE (and for paranoia also NA_BOT) explicitely set to zero. 2000-02-27 Bernd Kreimeier * unix/Makefile: added dmalloc in an attempt to get on the Z_Free bug. Futile. Despite stripping dmalloc debug token down to essentials, I get a (seemingly bogus or unrelated): debug-malloc library: dumping program, fatal error Error: possibly bad .c filename pointer (err 24) 2000-02-26 Bernd Kreimeier * qcommon/common.c: various debug builts to isolate the Z_Free bug. It reproducibly happens on some machines with SE_PACKET, but the packets themselves look thoroughly corrupted. 2000-02-21 Bernd Kreimeier * qcommon/common.c (Com_EventLoop): possible problem here, pointer does not get cleared. * unix/linux_glimp.c (InitSig): no signal handler. * common/common.c: dump in Com_Error for debug. 2000-02-17 Bernd Kreimeier * q3code: new dump from Zoid. Repeat tr_image.c fix. * unix/Makefile: added client/snd_adpcm.c (linkage errors). Later: added entire JPDIR and rules, for tr_image.c. Later: had to fix fules for game/ai_*.c files. Later: removed ui/ui_quit.o (n/a) Later: took out -mpentiumpro -march=pentiumpro * renderer/tr_image.c: windowism in #include path (see below). #include "..\jpeg-6\jpeglib.h" 1999-12-27 Bernd Kreimeier * Alpha: tried a dedicated server compile. Segfaults in ../qcommon/files.c:1682, a paksort function doing pointer fiddling. * Makefile.alpha: created. Note: want to take the SDL/Setup autoconf ASAP. * unix/unix_main.c: fixed __axp__ to __alpha__, guarded _FPU_SETCW. * qcommon/vm_alpha.c: dummy, created. * qcommon/vm_null.c: dummy, created. 1999-12-04 Bernd Kreimeier * renderer/tr_image.c: windowism in #include path. #include "..\jpeg-6\jpeglib.h" * Revision 1.11: from Zoid by e-mail. Note: threw away my playground copy, starting with the ZIP file. Zoid's using CVS now, but we can't remote access it. Thus did the "find . -name 'CVS' -exec rm -rf {} \;" and then track it as 3rd party source by * ChangeLog: created. Now starting to track Q3A source. --------- q3code log ---------------------------------------------