courtesy of configure. Some basic optimisations are added
unless optimisations are explicitly disabled. Extra
optimisations are added based on the cpu being built on.
module.so, and so weren't finding the modules (cunningly
hidden by the fact that old .so files weren't being
cleaned up by the makefile). Fixed that up good.
little more clearer
- The floating point exception when using the assembler
refreshers no longer occurs, which seems to be a
by-product of the recent auto* changes. If you can make
quake2 FPE again, I'd like to know. (Closes: #22)
- Added a configure test for ossaudio for NetBSD.
- Added a configure test for POSIX threads, from the GNU
autoconf macro repository.
- Fixed up some typos in the Rogue and Xatrix makefiles
are installed into the correct places.
- Added libltdl, the libtool dynamic loader to the code, as
the libraries are now under libtool control. Libtool and
libltdl are evil.
- Fixed various paths inside the code to point to PKGLIBDIR
for the game and refresher modules, and PKGDATADIR for the
game data.
- updated .gitignore
SVGAlib, X, GL, SDL, and GLIDE. Modified the
Makefile.am's to match the new configure variables.
Now people don't have to manually hack src/Makefile.am to
build their targets.
- Fixed a typo in src/qgl.h with the #define
- Added new automake generated files to bootstrap's clean
command.
- Added build requirements to HACKING, as we now need
automake 1.6.
- Removed note in README about the asm FPE bug, as I seem to
have inadvertently fixed this without meaning to.
- Commented out USE_ASM in the .S files -- in theory they're
only built when USE_ASM is enabled anyway...
- vid_so.c had some bad typecasts hidden away in some dead
code that became activated during this FPE crusade, and
though it's still dead, I thought I'd leave it as it is
now.
- Commented out some suspicious ifdefs in src/sys_dosa.S
- Fixed USE_ASM to be HAVE_MASM in src/ctf/q_shared.c
- something similar in src/baseq2/q_shared.c
- Added resolutions for the Sony Vaio Pocketbook and the
Apple Titanium Powerbook, as well as a note in HACKING for
people wishing to add their own resolutions. (Closes#33)
I have an idea involving querying the X server for supported
resolutions at runtime, or through some API that we can
use to do it portably (to SVGA/3dfx/Win32).
it to 1 to enable stencilbuffer shadows. (Turn gl_shadows
on first!) Based on MrG's tutorial on quakesrc.org, but
hacked a bit to work for GLX instead of Win32, and is a
bit more robust (won't die if it can't get a stencil
buffer).
- Updated TODO list to reflect this change.
VA_LIST_IS_ARRAY. Currently not needed, but I suspect we
may need them as we gain functionality; cf. dstring.c in
quakeforge.
- Updated configure.in to check for above tests.
- Updated TODO list noting the va_list audit.
that fixes gamma under GLX. His comments:
* This Linux Quake 2 patch allows to change OpenGL gamma
on the fly without vid_restart. It uses hardware gamma
where available (tested on XFree86 4.2.0 with DRI)
* This is for GLX only, SDL should have its own
implementation, based on its own gamma manipulation
functions
* This is implemented for fullscreen only. I'm not sure
if it shuld be implemented for windowed mode
* I've got the idea from MrG's BeefQuake for Windows, but
the implementation is completely mine
I can confirm that this works beautifully with XFree86
4.1, too.
project.
- Added HACKING for developer notes.
- Updated README to be more of a README.
- Updated TODO list.
- default don't build xatrix and rogue
- rolled back save game changes from 6 days ago, as they
suck. Save games are a complex beast and need someone to
redesign them so that they aren't tied to a particular
build. In the meantime, though, I'm not losing any sleep
over it, there's a note in the README about it, so time to
close this bug. (Closes: #15)
- commented out savegame check, which lets you use old
savegames. I'm not closing the bug, as I'm not convinced
it was this easy -- if the game data changes, then the
savegame data is invalidated, so in theory one should
check for game data version instead of binary version.
need to set dma.buffer to 0 so the sound devive gets mapped next time
around.
snd_dma.c:
register the commands after initializing the sound so that you don't
get errors about already existing commands if the sound init failed a
previous time around.
