The current GetVideoMode() function stops at the first mode which has any
dimensions smaller than the one asked, and gives the previous in the list.
If I ask 336x224 with this list:
768x480 768x240 640x400 640x200 384x480 384x240 320x400 320x200
SDL will give me 640x400, because 640x200 as height smaller than what I
asked.
However the best mode is the smaller which has both dimensions bigger
than the one asked (384x240 in my example).
This patch fixes this, plus it does not rely on a sorted video mode list.
From: Peter Naulls
Subject: RISC OS SDL patch
Some of the RISC OS audio code is no longer used do to the switch to
/dev/dsp, and can be removed as per the attached diff.
From: Jonathan Atkins
Subject: Re: [PATCH] *CRITICAL* 8bit direct RGB palette not being created
I think that SDL_AllocFormat should create the palette for all 8bit
surfaces. And when the RGBAmasks match the normal 3:3:2:0 we need to
apply the old behavior. If the mask doesn't match that, then we need
to make the right palette assuming the masks are valid (I don't think
we validate any masks for high color surfaces...so we wouldn't here)
Then there's always a palette available for the 8bit surfaces.
This restores the normal behavior and allows for masks to create
palettes automatically for odd masks even, which would be a neato
thing to have in there, as SDL never did this before.
OS report these are "horizontal scrollwheel" events, which confuses gaming
apps in several legitimate conditions. Now all scrollwheel events are made
to look vertical when passed to the app.
Patch by John Knottenbelt.
http://www.libsdl.org/pipermail/sdl/2005-March/067978.html
Jesse Barnes
Fri Jan 14 11:13:54 PST 2005
I noticed that on my ia64 machine when SDL_Quit was called, the machine would
hang in weird ways. It turned out to be caused by a machine check in the
memset() call near the top of FB_VideoQuit. Generally memset shouldn't be
used on I/O regions like the framebuffer or other I/O device memory (the
Linux kernel has special routines for dealing with I/O memory), so I changed
the #ifdef __powerpc__ to #if defined(__powerpc__) || defined(__ia64__) and
SDL_Quit now works properly. The change should probably be made
unconditional though since doing one byte at a time is the only way to be
portable, afaik.
From: Peter Naulls
Subject: RISC OS SDL Patches
Sam, I've attached a diff of the latest changes to libSDL for RISC OS
support. These changes are by Alan Buckley and myself.
The most significant of these are:
Optimised assembler blit rountines - I've attached the file
src/video/riscos/SDL_riscosASM.s which is needed for this.
Move to using /dev/dsp instead of its own audio implementation.
This means that src/audio/riscos/SDL_drenderer.c should be removed
Typo fixes. Mainly correct spelling of "RISC OS", but some from elsewhere
too.
From: Chris Taylor
Subject: Patch to put back dynamic OpenGL loading for MPW
I sent a patch a while ago that removes dynamic OpenGL loading for
Macintosh Programmer's Workshop. Dynamic loading DOES actually work
when an SDL program is built with MPW, it just has to be set up for it.
(Whoops!!) This is the ideal way to get OpenGL extensions to work,
which D2X uses quite a few of.
This patch puts dynamic loading back in SDL for Mac OS 9. It applies to
current CVS. I noticed that two members need to be set when
DrawSprocket is used.
From: Ville [snip]
Subject: [PATCH] SDL/DirectFB: remove Matrox CRTC2 flicker filter
Hi,
This patch removes the flicker filter option from the DirectFB backend's
Matrox CRTC2 code in SDL. I will be removing the option from DirectFB
(because it doesn't actually work) and that would cause the SDL code to
fail without this fix.
I was going to send this to some SDL list directly but libsdl.org is down
so I'm not sure what if any lists there are. Thomas Jarosch (the guy who
wrote the code) said you would accept SDL patches. Let me know if I should
send this somewhere else.
-- Ville Syrj�l� syrjala _at sci.fi http://www.sci.fi/~syrjala/
