Casts to (char *) will disable strict aliasing when the compiler sees it.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401164
just be flat out missing from glibc, so force use of previous gettimeofday()
behaviour for now.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401161
From: Tommi Kyntola <tommi.kyntola@ray.fi>
To: sdl@libsdl.org
Subject: [SDL] [RFC] get_ticks broken on linux
It uses gettimeofday to calculate the timedifference.
Gettimeofday returns current time which is seldom monotonous.
This breaks SDL timer subsystem. (time callbacks and all that
get borked when the time difference ms is suddenly ~ 2^32)
I posted a message about this earlier but got no response.
Some thoughts on this matter would be appreciated.
(Or even an explanation for the lack of interest.)
A patch below would use the posix timers that have been around
since posix 93 and do provide a good source of monotonous time
on linux boxes (and on few others too).
The following patch is also availabe at:
http://www.hut.fi/u/tkyntola/SDL-1.2.9-clockfix.patch
It's against 1.2.9, but I can easily rediffit against
the cvs branch is needed.
cheers,
Tommi Kyntola tommi.kyntola@ray.fi
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401160
From: =?ISO-8859-1?Q?Martin_Storsj=F6?= <martin@martin.st>
To: sdl@libsdl.org
Subject: [SDL] [PATCH] Use nanosleep on OS X
Hi,
The current version of SDL_Delay on OS X seems to always sleep at least 10
msec. OS X has nanosleep(), which performs shorter sleeps well. The
attached patch makes it use that one instead of select() as currently.
// Martin
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401159
if SDL is built with a non-cdecl calling convention, and it's just generally
bad practice anyhow.
Now programs explicitly call SDL_Quit() where appropriate, wrap SDL_Quit() in
a cdecl function where it can't be avoided, and rely on the parachute where
a crash might have hit the atexit() before (these ARE test programs, after
all!).
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401154
http://bugs.gentoo.org/show_bug.cgi?id=104437
I believe the patch is from Harald van Dijk (comment #16 at the above url).
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401151
To: "'A list for developers using the SDL library. \(includesSDL-announce\)'" <sdl@libsdl.org>
Date: Mon, 19 Sep 2005 18:59:43 -0400
Subject: [SDL] [patch] Volume multiplier bug in
The volume multiplier in SDL_MixAudio_MMX_S8() is formed from the wrong
register.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401150
From: Jon Daniel <forcemaster@gmx.net>
To: sdl@libsdl.org
Subject: [SDL] SDL_SemWaitTimeout patch for BeOS
The current SDL-1.2.9/src/thread/beos/SDL_syssem.c returns -1 if the
acquire_sem timeouts which according to the DocWiki is not correct.
I've added support for the B_TIMED_OUT and B_WOULD_BLOCK in the switch.
Jon Daniel
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401149
windowclass when shutting down the video subsystem...this allows you to
safely unload/reload the SDL shared library on Windows between
initializations.
Discussion is here:
http://www.devolution.com/pipermail/sdl/2005-February/067424.html
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401148
From: =?ISO-8859-1?Q?Martin_Storsj=F6?= <martin@martin.st>
To: sdl@libsdl.org
Subject: [SDL] [PATCH] Switching between fullscreen/windowed on OS X
Hi,
When switching from fullscreen to windowed mode for opengl applications,
the current version of SDL sets an harmless error, regarding usage of a
NULL semaphore. This is due to code which tries to shut down a blitting
thread, which is only started for double buffered 2d video modes, not for
opengl. The attached patch fixes this.
// Martin
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401147
requested this effort on the mailing list, apparently because of binary
compatibility issues between 10.4 and earlier systems (or gcc4 and earlier
compilers?).
Works fine with SDL12/test/testcdrom.c, with this command line:
./testcdrom -status -list -play -sleep 5000 -pause -sleep 3000 -resume \
-sleep 5000 -stop -sleep 3000 -play -sleep 3000 -stop \
-sleep 3000 -eject
Unix Makefiles work, XCode project still need updating for new filenames.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401146
From: Mike Frysinger
Subject: [SDL] [patch] make sure autotools worked
the current autogen.sh runs a bunch of autotools without actually checking the
results ... find attached a patch to add some simple sanity checks
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401145
specifically. Patch by Anders F Bj�rklund <afb@algonet.se>.
--ryan.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401144
--ryan.
Date: Sun, 19 Jun 2005 15:53:22 -0700
From: Joshua Oreman <oremanj@gmail.com>
To: sdl@libsdl.org
Subject: [SDL] [PATCH] iPod framebuffer video driver
Hi SDL-list,
I've been working on a port of SDL to iPodLinux
(http://www.ipodlinux.org). I've created a patch for both the
standard 2-bit iPod screen (using an unchangeable palette) and the
16-bit iPod photo. The patch is attached, against version 1.2.8.
I've created two pages on the iPodLinux wiki about this patch:
http://www.ipodlinux.org/Building_SDL and
http://www.ipodlinux.org/SDL_Programming. See those pages if you're
curious.
Comments? Questions? Is this something that might be able to get into SDL 1.2.9?
Thanks for your feedback!
-- Josh
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401143
--ryan.
Date: Sun, 14 Aug 2005 23:06:40 -0500
From: Shawn Walker <binarycrusader@gmail.com>
To: sdl@libsdl.org
Subject: [SDL] [PATCH] Audio Detection Bug
When using the OSS commercial drivers under Solaris 10, SDL will not
properly initialise OSS audio support (dsp) if /dev/sound exists.
Under Solaris (as far as I understand) /dev/sound is provided as a
means of accessing a BSD style audio device, not the OSS device.
SDL assumes that if /dev/sound exists, then it must be running on a
Linux 2.4 system and should make the dsp device path /dev/sound/dsp.
This is wrong. When using the OSS commercial drivers under Solaris,
the dsp device is always referenced as /dev/dsp normally.
My proposed fix is to stat the dsp device in /dev/sound to make sure
it exists, before assuming /dev/sound/dsp as the audio device:
http://icculus.org/~eviltypeguy/SDL_audiodev.patch
I'm sure there may be a better way to do it, but the above patch is
what worked for me.
--=20
Shawn Walker, Software and Systems Analyst
binarycrusader@gmail.com - http://binarycrusader.blogspot.com/
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401141
--ryan.
From: Mike Frysinger <vapier@gentoo.org>
To: sdl@libsdl.org
Date: Thu, 1 Sep 2005 20:25:01 -0400
Subject: [SDL] [patch] add support for arm/thumb and superh to endian
find attached a patch to add support for detecting endian on superh and
arm/thumb ... also, ive incorporated a patch from Fedora which will gather
the correct endian on a linux host from the system endian.h instead of just
trying to maintain an ever-growing list of architectures
-mike
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401140
--ryan.
From: Mike Frysinger <vapier@gentoo.org>
To: sdl@libsdl.org
Subject: Re: [SDL] sdl, sdl-mixer not compiling with uclibc
Date: Wed, 7 Sep 2005 19:17:49 -0400
in this case actually, the error is in libsdl
the configure check for dlvsym presence is pretty pointless, it currently
always detects it
AC_TRY_COMPILE([
#include <stdio.h>
#define __USE_GNU
#include <dlfcn.h>
],[
dlvsym(NULL,"","");
],[
use_dlvsym=yes
])
since it's merely compiling, the warning about dlvsym being implicitly
declared doesnt trigger a failure
find attached a patch to fix this
-mike
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401139
--ryan.
From: =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb@algonet.se>
Subject: Re: [SDL] Updated Mac patch
Date: Tue, 6 Sep 2005 15:21:27 +0200
To: A list for developers using the SDL library <sdl@libsdl.org>
Earlier, I wrote:
> Updated the previous Mac patch to disable Carbon by default.
> Also "fixed" the SDL.spec again, so that it builds on Darwin.
>
> http://www.algonet.se/~afb/SDL-1.2.9-mac.patch
> Also applied fine to SDL12 CVS, when I tried it.
>
> Haven't completed any new packaging or projects for Xcode/PB,
> but it seems to build and install fine here (in development).
Tested the new patch to build with old CodeWarrior and MPW,
and it seems it needed some hacks with those old headers...
Just in case you want to support the archeological versions -
here is a small add-on to the above patch, to fix those...
http://www.algonet.se/~afb/SDL-1.2.9-classic.patch
I couldn't get the old CW5 projects to build without a few
modifications - such as deleting the stray old header in:
"CWprojects/Support/Carbon/Include/ConditionalMacros.h" ?
But I updated both projects to CW6 too and built for Carbon,
and it ran all of the Mac test projects without any problems.
The MPW file seems to have compiled, with a small order change.
As long as you're still shipping the CWProjects and MPWmake
with the download, they should probably be updated/fixed ?
(another "solution" would of course be to just delete them)
I'll post my new projects along with the new Xcode projects
later on, along with XML exports of the various .mcp files.
(CW5 builds for Classic / "PPC", and CW6 builds for Carbon)
It'll be packaged as a part of the next SpriteWorld X release...
http://spriteworldx.sourceforge.net/ [Classic/Carbon/Win/X11]
--anders
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401138
To: SDL Developers <sdl@libsdl.org>
From: =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb@algonet.se>
Date: Mon, 30 May 2005 23:29:04 +0200
Subject: [SDL] Mac OS X Video Drivers [patch]
I've updated/added the Carbon and X11 video drivers
to the Mac OS X port of SDL 1.2 (the CVS version),
and made the Cocoa driver and runtime *optional*.
The default is still Cocoa, and the "Quartz" driver.
But you can now also use "toolbox" for Carbon, and
"x11" for running with Apple's (or other) X11 server:
export SDL_VIDEODRIVER=x11
export SDL_VIDEO_GL_DRIVER=/usr/X11R6/lib/libGL.dylib
It also checks if the frameworks are available, by a:
#include <Carbon/Carbon.h> or #import <Cocoa/Cocoa.h>
(this should make it configure on plain Darwin as well?)
Here are the new configure targets:
--enable-video-cocoa use Cocoa/Quartz video driver default=yes
--enable-video-carbon use Carbon/QuickDraw video driver default=yes
--enable-video-x11 use X11 video driver default=no
./configure --enable-video-cocoa --enable-video-carbon
--enable-video-x11 \
--x-includes=/usr/X11R6/include --x-libraries=/usr/X11R6/lib
The Carbon version is just an updated version of the old
SDL driver for Mac OS 9, and could probably be improved...
(but it does work, including the Carbon version of SDLmain)
If you disable cocoa, you can run with -framework Carbon only,
and the C version of SDL_main.c. And if you disable carbon too,
you can still use the X11 version which doesn't require SDLmain.
I updated the DrawSprocket version, but did not include it.
(no blitters or VRAM GWorlds etc. available on OS X anyway)
Besides for Mac OS 9, I don't think there's any use for it ?
And note that any performance on Mac OS X equals OpenGL anyway...
You can get "fair" software SDL results on captured CG displays,
but for decent frame rates you need to be using GL for rendering.
Finally, here is the patch itself:
http://www.algonet.se/~afb/SDL-12CVS-macvideo.patch
--anders
PS. It says "video", but as usual it applies to mouse/keyboard too.
------
To: A list for developers using the SDL library <sdl@libsdl.org>
From: =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb@algonet.se>
Date: Sun, 4 Sep 2005 10:02:15 +0200
Subject: [SDL] Updated Mac patch
Updated the previous Mac patch to disable Carbon by default.
Also "fixed" the SDL.spec again, so that it builds on Darwin.
http://www.algonet.se/~afb/SDL-1.2.9-mac.patch
Also applied fine to SDL12 CVS, when I tried it.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401136
From: "Ryan C. Gordon"
Subject: [Fwd: SDL patch: make usage of rpath optional]
Here's a small patch against current SDL CVS that makes usage of rpath
optional, by passing --disable-rpath to configure. This comes in handy
when redistributing SDL -- the rpath setting prevents the lib being
loaded if it's not in the rpath, which makes redistributing in packages
to be installed by non-root users pretty much useless.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401128
"I'm on Mac OS 10.3.9 with a CVS SDL 1.2.9.
My understanding is that SDL_CDResume is supposed to resume play after
calling SDL_CDPlay. It doesn't on my system. It returns 0 but nothing happens.
Any ideas?
Thanks.
Mike Benfield"
--ryan.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401127
app as an openFile() message, so we have to make sure we were launched from
the Finder before accepting these as drag'n'drop documents, or they will just
duplicate what's already in argc/argv.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401126
From: Jiri Svoboda
Subject: [SDL] signal handling bug
I encountered the following bug:
SDL doesn't reset signal handlers for SIGTERM and SIGINT, after calling SDL_Quit these remain hooked to the handler in SDL_quit.c, being translated into SDL_QUIT events.
Consequently an application that issues a SDL_Quit and remains running will ignore any SIGTERM or SIGINT., and specifically CTRL-C presses.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401124
"Just a request about test subdirectory: the autotools files
(config.guess, config.sub, etc...) are a bit outdated, and I have to
update them for my cross-compilation party. So it would be nice to
have these updated.
...
Just removing these files works, because then configure will use
those of SDL's topdir. Could someone else check this, just to be sure?
i.e. remove config.guess, config.sub, install-sh and mkinstalldirs from
test subdir."
I tested on Linux and MacOSX...seems to be safe, so I'm nuking the files.
--ryan.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401123
--ryan.
To: sdl@libsdl.org
From: Rainer Deyke <rainerd@eldwood.com>
Date: Tue, 16 Aug 2005 01:08:18 -0600
Subject: [SDL] Bug report: SDL_CreateRGBSurface with SDL_HWSURFACE crashes
If SDL is in full-screen mode with a hardware video surface on OS X,
SDL_CreateRGBSurface with SDL_HWSURFACE crashes. The crash occurs on
line 109 of SDL_Surface.c. This was tested on OS X 10.3.9 with both SDL
1.2.8 and the latest CVS. Here is a small C++ program that demonstrates
the problem:
#include "SDL.h"
#include <stdio.h>
namespace {
void wait_for_key()
{
SDL_Event e;
printf("%d\n", SDL_GetAppState());
while (SDL_WaitEvent(&e)) {
if (e.type == SDL_KEYDOWN || e.type == SDL_QUIT) return;
}
}
}
int main(int, char *[])
{
SDL_Init(SDL_INIT_VIDEO);
SDL_Surface *screen
= SDL_SetVideoMode(640, 480, 32, SDL_FULLSCREEN | SDL_HWSURFACE);
SDL_Surface *s = SDL_CreateRGBSurface(SDL_HWSURFACE, 640, 480, 32,
screen->format->Rmask, screen->format->Gmask,
screen->format->Bmask, screen->format->Amask);
wait_for_key();
if (s) SDL_FreeSurface(s);
SDL_Quit();
printf("Success!\n");
return 0;
}
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401121
Patch by me to address this comment on the SDL mailing list:
"There appears to be a serious bug with SDL_GetAppState on OS X (10.3.9). When
first running in windowed mode and then switching to full screen mode by
calling SDL_SetVideoMode, SDL_GetAppState returns 4 (SDL_APPACTIVE) instead of
7 (SDL_APPMOUSEFOCUS | SDL_APPINPUTFOCUS | SDL_APPACTIVE). However, the SDL
application clearly does have the keyboard focus, since it is able to receive
keyboard events."
--ryan.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401120
From: Matt L <prometheus.uw@gmail.com>
To: "Ryan C. Gordon" <icculus@clutteredmind.org>
Subject: SDL Patch
Hio,
I saw your last call on the mailing list. Here's a patch which I
submitted about two weeks ago which hasn't made it in.
In the current sdl.m4, there's a bug where if your configure.ac, you
have AC_LANG(C++) specified, it won't properly compile and link the SDL
test program when you run the configure script. This is because only the
default CFLAGS is overriden in sdl.m4, and in the patch below, I've
fixed it so it overrides CXXFLAGS as well, allowing it to work with g++.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401119
To: sdl@libsdl.org
Date: Sat, 9 Jul 2005 13:59:33 -0400
Subject: [PATCH] fix locking in src/timer/SYS_timer.c
SDL_SetTimer has a typo in CVS. This code was added since 1.2.8. The
result is that the SDL_timer_mutex is locked twice and never unlocked,
breaking systems that use a threaded timer implementation.
-jim
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401118
document associated with the app, the document(s) are passed to SDL_main()
as if they were command line arguments. Otherwise, the command line is always
empty and there is no way for the app to recover this information.
--HG--
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/trunk%401114
