The software renderer has been re-routed to use the framebuffer interface, which makes it possible to have software rendering available even on simple ports.
Vittorio Giovara 2011-02-01 02:21:50 PST
with the attached patch, the opengles context will always use the maximum
screensize available; this is particularly useful for supporting retina display
on latest iphone.
please note: Apple documentation warns that using the "upscaled" gl context
actually uses more memory and bandwitdh, so it might be worth to let the user
decide whether to disable it or not, either with a flag or a sdl function...
kwm@rainbow-runner.nl 2011-01-30 06:28:27 PST
Created attachment 562 [details]
Build fix with clang.
When building sdl 1.2.14 with the Clang compiler http://clang.llvm.org .
The build fails in src/video/mmx.h with the following error:
--------------------------------------------------
./src/video/SDL_RLEaccel.c:831:5: error: invalid operand for instruction
CHOOSE_BLIT(RLECLIPBLIT, alpha, fmt);
^
./src/video/SDL_RLEaccel.c:831:17: note: instantiated from:
CHOOSE_BLIT(RLECLIPBLIT, alpha, fmt);
^
./src/video/SDL_RLEaccel.c:831:5: note: instantiated from:
CHOOSE_BLIT(RLECLIPBLIT, alpha, fmt);
^
./src/video/SDL_RLEaccel.c:647:23: note: instantiated from:
blitter(2, Uint8, ALPHA_BLIT16_565MMX); \
^
./src/video/SDL_RLEaccel.c:282:4: note: instantiated from:
movq_r2m(mm3, *dstp); \
^
In file included from ./src/video/SDL_RLEaccel.c:99:
./src/video/mmx.h:379:28: note: instantiated from:
#define movq_r2m(reg, var) mmx_r2m(movq, reg, var)
^
<scratch space>:192:1: note: instantiated from:
"movq"
^
<inline asm>:1:2: note: instantiated into assembly here
movq %mm3, %dx
^
--------------------------------------------------
According to the clang developers this is a invalid inline assembly.
Using the attached patch from the last commit in the below bug report fixes the
compile.
More details from: http://llvm.org/bugs/show_bug.cgi?id=6730
Joseba García Echebarria 2010-12-15 01:55:22 PST
I believe the crash is caused by a check not being performed on wether an
SDL_Touch element is NULL before using it in the SDL_SendTouchMotion function
in src/events/SDL_touch.c around line 400.
Judging from the rest of the code, there's a missing
if (!touch) {
return 0;
}
before using "touch" as SDL_GetFinger(), SDL_GetFingerIndexId() use
touch->num_fingers without checking.
I can attach a patch if you like. It seems pretty straightforward, though.
I have yet to discover why touch is being returned as NULL as this error is
only triggered when an actual gesture has been performed, maybe something
related to SDL_AddTouch()?
Common.c now destroys the SDL_Window upon a SDL_WINDOWEVENT_CLOSE event to ensure that all windows get closed properly and the new code to handle the last window closes gets executed.
The new timer model is formalized as using a separate thread to handle timer callbacks. This was the case on almost every platform before, but it's now a requirement, and simplifies the implementation and makes it perform consistently across platforms.
Goals:
* Minimize timer thread blocking
* Dispatch timers as accurately as possible
* SDL_AddTimer() and SDL_RemoveTimer() are completely threadsafe
* SDL_RemoveTimer() doesn't crash with a timer that's expired or removed
the pthread implementation of SDL_SemWaitTimeout() uses busy waiting, while
pthread's sem_timedwait() does work. Attached are patches that make use of it
Markus Rathgeb 2011-01-23 14:34:23 PST
With kernel 2.6.31 the struct input_absinfo defined in linux/input.h changed.
A field "__s32 resolution" was added at the end of the struct.
Because the macro EVIOCGABS(abs) is using the struct input_absinfo, it would be
better (IMHO) to change the declaration of variable values to
"int values[sizeof(struct input_absinfo) / sizeof(int)];" or using "struct
input_absinfo" directly.
Daniel Ellis 2010-06-25 15:20:31 PDT
SDL based applications sometimes display the wrong application name in the
Sound Preferences dialog when using pulseaudio.
I can see from the code that the SDL pulse module is initiating a new pulse
audio context and passing an application name using the function
get_progname().
The get_progname() function returns the name of the current process. However,
the process name is often not a suitable name to use. For example, the OpenShot
video editor is a python application, and so "python" is displayed in the Sound
Preferences window (see Bug #596504), when it should be displaying "OpenShot".
PulseAudio allows applications to specify the application name, either at the
time the context is created (as SDL does currently), or by special environment
variables (see http://www.pulseaudio.org/wiki/ApplicationProperties). If no
name is specified, then pulseaudio will determine the name based on the
process.
If you specify the application name when initiating the pulseaudio context,
then that will override any application name specified using an environment
variable.
As libsdl is a library, I believe the solution is for libsdl to not specify any
application name when initiating a pulseaudio context, which will enable
applications to specify the application name using environment variables. In
the case that the applications do not specify anything, pulseaudio will fall
back to using the process name anyway.
The attached patch removes the get_progname() function and passes NULL as the
application name when creating the pulseaudio context, which fixes the issue.
You can also do this through the native API:
UIWindow *window = [[UIApplication sharedApplication] keyWindow];
Also needed to name the union for events and window info.
Vittorio Giovara 2010-07-16 19:09:28 PDT
i was reading SDL_renderer_gles and i noticed that every time we there
is some gl call the gl state is modified with a couple of
glEnableClientState()/glDisableClientState.
While this is completely fine for desktops systems, this is a major
performace kill on mobile devices, right where opengles is
implemented.
Normal practice in this case is to update the glstate once, keep it
always the same and disable/enable other states only in very special
occasions.
On the web there's plenty of documentation (on the top of my head
http://developer.apple.com/iphone/library/documentation/3DDrawing/Conceptual/OpenGLES_ProgrammingGuide/Performance/Performance.html#//apple_ref/doc/uid/TP40008793-CH105-SW5
) and i personally tried this.
I modified my code and got a 10 fps boost, then modified SDL_render_gles and
shifted from 40 fps to 50 fps alone -- considering that i started from ~30fps i
got an 80% performance increase with this technique.
I have attached a dif of my changes, hope that it will be included in
mainstream.
pelya 2010-07-21 04:54:41 PDT
GLES_UpdateTexture() ignores pitch value, because of that some textures created
with SDL_CreateTextureFromSurface() are drawn incorrectly, especially if image
width is not multiplier of 2.
Vittorio Giovara 2010-07-17 19:21:36 PDT
fix the double free error
in SDL_uikitview the variable 'textfield' is initialialized and set to
autorelease.
however in the dealloc method a second [release] is sent.
If the textfield has not been set to nil before (with a viewDidUnload for
example) this can lead to awful hard-to-find crashes when the SDL code
terminates.
the error message is
-[textfield release] message sent to deallocated instance 0x4e5fa90
the fix is simple, just release the object as soon as it is added to the
subview, see attached patch
The are no event handlers for the middle mouse button, right mouse buttons, and
mouse wheel in the latest HG revision 4636. It has been like this for 3 months.
I made a patch for this, though I'm not sure if the Xbutton or mouse wheel code
is "correct" by your standards.
in SDL 1.3 (revision 5508 from SVN), the method used to calculate the
bits per pixel from a “int format” differ between “SDL_ListModes” (which
always uses the “SDL_BITSPERPIXEL” macro) and “SDL_PixelFormatEnumTo-
Masks” (which uses either “SDL_BITSPERPIXEL” or “SDL_BYTESPERPIXEL * 8”,
depending on the value of “SDL_BYTESPERPIXEL”).
Because the values are later compared in “SDL_ListModes” this may lead
to some valid video modes not being returned. In my case the only mode
returned by “SDL_GetNumDisplayModes” was dismissed and NULL was
returned. (This led to the calling application sticking its head in the
sand.)
The attached patch copies the method used within “SDL_PixelFormatEnumTo-
Masks” to “SDL_ListModes”. This solved the problem for me though I don't
fully understand the method used by “SDL_PixelFormatEnumToMasks”.
