Ludwig Nussel      2009-10-18 05:34:18 PDT

src/joystick/linux/SDL_sysjoystick.c has some problems:

- test_bit() might break with strict aliasing
- test_bit() assumes array is Uint32 but its actually "unsigned long"
  on 64bit systems sizeof(long) != sizeof(Uint32).
- the keybit array is too small
- the arrays are unitialized so the number of
  detected buttons is quite random

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404109

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404105

This was a silly bug. :)

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404104

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404101

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404100

For some people setting the period size works better (and is what SDL 1.2.13 did), but for most people it's the same or worse.  You can use an environment variable to pick which one you want.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404099

actually, please don't apply this upstream.  i need to think about
a better solution..

thanks.

.mrg.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404098

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404097

Fix assertion triggered in pulsaudio by returning "unknown" instead
of NULL pointer when guessing application name.

- Use getprogname on NetBSD as it wont return NULL

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404096

Look for libGL.so, not any particular libGL.so.<X>, on NetBSD.
(the comment inline explains why)

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404095

Apply patch to first unlock the CD before trying to eject it. Since
SDL doesn't seem to have lock/unlock primitives, this looks like the
right thing to do.

Fixes audio/tcd's eject function too.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404094

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404093

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404092

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404091

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404090

I am using a headset on my monitor as primary sound output device. From time to
time it happens that I switch off my monitor when I leave my computer and
switch it on when I come back.
When doing so while wesnoth is running I get no sound back when I switch on my
monitor. There is probably good reason for that. Therefore I wanted to exit and
restart wesnoth. Unfortunately, wesnoth always hangs in this situation. It
looks like wesnoth can not be exited when the primary sound device vanished.
I created a backtrace of this situation (please see attachment).

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404089

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404088

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404084

Use the plug: ALSA devices for surround sound formats, which have automatic sample format/rate conversion.
Based on feedback from the ALSA development mailing list:
http://mailman.alsa-project.org/pipermail/alsa-devel/2009-October/022267.html

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404081

Switched from setting the period size and count to setting the buffer size directly, based on feedback from the ALSA development list:
http://mailman.alsa-project.org/pipermail/alsa-devel/2009-October/022267.html

This has the nice side effect of reducing latency on my SBLive! card.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404080

ALSA's idea of a "frame" is the same as SDL's idea of a "sample".  The frame is a single sample on a single channel, and we've defined the frames for each channel as being interleaved.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404079

Passed a meaningful value to snd_pcm_sw_params_set_start_threshold()

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404078

On recent distributions ALSA pipes streams through PulseAudio anyway, so using PulseAudio first when it's available will improve latency and reduce ALSA configuration quirks.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404077

Apparently ALSA says "frames" but it means samples...you don't split this into
 groups by number of channels.

The adventure continues.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404076

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404072

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404067

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404066

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404065

While trying to build the SDLMain.m included with SDL 1.2.14, with
#define SDL_USE_NIB_FILE 1:

/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:
In function '-[SDLMain fixMenu:withAppName:]':
/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:122:
warning: 'sizeToFit' is deprecated (declared at
/Developer/SDKs/MacOSX10.6.sdk/System/Library/Frameworks/AppKit.framework/Headers/NSMenu.h:281)
/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:
In function 'main':
/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:376:
warning: 'poseAsClass:' is deprecated (declared at
/Developer/SDKs/MacOSX10.6.sdk/System/Library/Frameworks/Foundation.framework/Headers/NSObject.h:127)
/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:376:
error: 'poseAsClass:' is unavailable (declared at
/Developer/SDKs/MacOSX10.6.sdk/System/Library/Frameworks/Foundation.framework/Headers/NSObject.h:127)
/Users/steven/Development/darwinia/targets/macosx/Darwinia/SDLMain.m:377:
warning: passing argument 2 of 'NSApplicationMain' from incompatible
pointer type



Eric Wing to Sam

I don't have time today to look at this in detail, but the problem is definitely the poseAsClass: method.
This was deprecated in Obj-C 2.0 and not retained in 64-bit.

I've never used this method and it has always been limited to esoteric uses. I think this is why Apple wanted to dump it (among complicating some other things they do). I have read about others getting bit by this when migrating. Long story short, there really isn't a migration path for this method. The question that then must be asked is why are we using it (what does it accomplish), and then figure out the 'proper' way of accomplishing that.

Glancing at SDLMain.m, it's not obvious to me why it is there or what it is really accomplishing. My only speculation is that NSApplicationMain hardcodes something to look for NSApplication and a subclass (SDLApplication) fails for some reason (assuming that the original coder did this for good reason).

Three thoughts come to mind.
1) The Info.plist has properties to control things related to the startup class and nib.
NSPrincipalClass, NSMainNibFile

Maybe principle class needs to be SDLApplication and we can delete the poseAs

2) I was told that 10.6 introduced new APIs to make programatic NIB wrangling and avoidance easier. Unfortunately, I don't know the specifics.


3) Instead of subclassing NSApplication in SDLMain.m, maybe we can just add a category. It looks like the following is the only thing that is done (quick glance):

@interface SDLApplication : NSApplication
@end

@implementation SDLApplication
/* Invoked from the Quit menu item */
- (void)terminate:(id)sender
{
   /* Post a SDL_QUIT event */
   SDL_Event event;
   event.type = SDL_QUIT;
   SDL_PushEvent(&event);
}
@end


So instead, we change this to: (warning written in mail and untested)

@interface NSApplication (SDLApplication)
- (void) terminate:(id)sender;
@end

@implementation NSApplication (SDLApplication)
/* Invoked from the Quit menu item */
- (void)terminate:(id)sender
{
   /* Post a SDL_QUIT event */
   SDL_Event event;
   event.type = SDL_QUIT;
   SDL_PushEvent(&event);
}
@end


Then everywhere SDLApplication is used, we change it to NSApplication (and remove the poseAsClass line).

Perhaps you could ask the bug reporter to try this solution (#3).
And if that fails, maybe try #1.


-Eric


Steven Noonan to Sam

The suggested change (diff below) seems to work fine.

- Steven

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404064

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404060

  Hopefully fixes Bugzilla #649.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404059

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404057

  Hopefully fixes Bugzilla #650.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404056

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404043

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404042

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404041

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404038

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404037

 Ozkan Sezer      2009-10-10 01:45:27 PDT

sdlenvvars.html doesn't list pulse among the supported values for
SDL_AUDIODRIVER var.  Attached a quick patch to remedy that, hopefully placing
the entry in the right order.  Regards.

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404036

--HG--
branch : SDL-1.2
extra : convert_revision : svn%3Ac70aab31-4412-0410-b14c-859654838e24/branches/SDL-1.2%404035