From: Darrell Walisser
Subject: Yet another OS X cursor bug
The synopsis:
1. Call SDL_ShowCursor(0);
2. Call SDL_SetVideoMode();
3. Call SDL_GetEvent();
3. Call SDL_ShowCursor(1);
The result: Sometimes the cursor doesn't come back! Ack! Oddly enough,
it does come back when mousing over the dock or clicking in the menu
bar. But that's besides the point.
The reason why this is happening is a flaw in the handling of
activation/deactivation events. The short explanation is that the
HideCursor() and ShowCursor() calls must be balanced, but if the cursor
was initially hidden, HideCursor() was called again on the activate
event - so now the next ShowCursor() fails (as does the next, and the
next, for some reason).
So, here's the patch. All it does is keep track of the
HideCursor()/ShowCursor() calls so that they will always be balanced.
From: Stephen Anthony <stephena@roadrunner.nf.net>
Subject: [SDL] First patch concerning 4.3 and refresh rates
OK, here's my first draft of the patch for the above subject.
A short explanation:
X 4.3 introduces many more modelines than older versions. This would be
fine, except it introduces many modes with the *same* resolution but
different refresh rates. And SDL won't necessarily pick the one with the
highest refresh rate.
So this patch restores SDL to X 4.2 functionality. That is, there is only
ever one refresh rate *per* resolution, and it is the highest possible.
This functionality can be totally disabled by using the environment
variable 'SDL_VIDEO_X11_USE_ALL_MODES' set equal to 1.
From: "Matthew N. Dodd"
Subject: SDL patch: FreeBSD joystick support.
This patch deals with the recent changes in FreeBSD.
We're making an effort not to diverge our libusbhid from NetBSD's.
From: Patrice Mandin
Subject: [SDL][PATCH] 2 patches for sdl
Here are 2 patches for SDL:
- One is to put the dummy video drivers at the end of the
video drivers list. It gave me problems, when
SDL_VIDEODRIVER is not set, and the dummy driver is used
instead of the platform's driver, just because it is
always available. So the dummy driver must always be at
the end of the list. I suppose picogui and dc video
drivers also don't work.
From: Denis Oliver Kropp
Subject: Makefile.am fix
I fixed the Makefile.am which lacked many defines.
With this patch SDL on DirectFB doesn't crash on
startup immediately.
From: Ragnvald "Despair" Maartmann-Moe IV
Subject: SDL x11 video mode selection bug
Hi, I've found an oddity with video mode selection. I recently added a bunch of
oddball video modes to my XF86Config, to support weird resolutions xine &
mplayer need if I don't want to beat my cpu up with video scaling.
Since adding them, SDL started picking the biggest video mode it could find
that matched the height constraint. Getting a narrow vertical strip of
QuakeForge (I code for the project, so needless to say it's my favorite test)
in the middle of an 800x480 screen, instead of the 640x480 I asked for
annoys me tremendously.
So here's a patch that tries a bit harder to get an exact match. I didn't
touch the XiG section, since I can't test that, but I'd bet a similar patch
would prevent similar problems there.
--Ragnvald "Despair" Maartmann-Moe IV
From: Patrice Mandin
Subject: [SDL] Small patches for SDL
- One for SDL, which removes check for SDL_VIDEODRIVER in the Atari video
drivers. It is unnecessary 'cause it is done in SDL_video.c
From: "Mike Gorchak"
Subject: All QNX patches
whole patches concerning QNX. Almost all code has been rewritten by Julian
and me. Added initial support for hw overlays in QNX and many many others
fixes.
P.S. This patches has been reviewed by Dave Rempel from QSSL and included in
SDL 1.2.5 distribution, which coming on 3rd party CD for newest 6.2.1
version of QNX, which will be available soon.
From: "Mike Gorchak"
added help and options -mono to force monochrome RGB2YUB conversion and
-lum <percent> - luminance of image during conversion. Also rewritten code
which moves overlay into window. Now it depends on window size.
Also fully rewritten -scale option, image now scaling from 50% from center
of screen until border of window is reached - it really tests scaler, old
-scale test doesn't test downscale.
