Icecast dev - Aug 2004 - timing_sleep malfunctioning under MinGW

Hi,
I have successfully gotten the libshout-2.0 release to compile under
gcc/MingGW ('gcc -mno-cygwin', version 'gcc (GCC) 3.3.1 (cygming
special)'), after applying Leigh Smith's patches as posted on this list
(http://www.xiph.org/archives/icecast-dev/0660.html). I have then
generated a dll and wrapped it in JNI using the stubs from libshout-java
(http://benow.ca/index?page=project&project=libshout-java). So, I have
libshout basically working through Java on Windows XP.

I think everything is working ok, except shout_sync. This call returns
immediately under the compilation setup above, instead of actually
sleeping, screwing up the audio streams. This occurs in exactly the same
way when I compile the ezstream utility off of the libraries as when I'm
running libshout through Java, so I don't think it's an artifact of the
JNI process.

Looking at timing_sleep, the only relevant call there is select(), so
I'm guessing the problem is in the invocation there, but I can't figure
out what would fix it -- I've been programming Java too long, I'm just
glad I made it this far :)

I modified the JNI wrappers to expose shout_delay and have been
handling sleeping on my own in Java, and everything seems to work
perfectly. So this isn't a debilitating issue for me, but I'm hoping
someone on the list will recognize what the fix would be and can get
shout_sync working at some point under MinGW. Thanks!

-alan

PS. under regular cygwin this problem doesn't occur; I need to go ming
for JNI.

Of course, only after posting to the list do I peek in the subversion
repository and notice that some Win32-centric patches to timing have
already been made. I'll try them out, sorry for the spam...

-alan

Alan Newberger wrote:
> Hi,
>     I have successfully gotten the libshout-2.0 release to compile under
> gcc/MingGW ('gcc -mno-cygwin', version 'gcc (GCC) 3.3.1
(cygming
> special)'), after applying Leigh Smith's patches as posted on this
list
> (http://www.xiph.org/archives/icecast-dev/0660.html). I have then
> generated a dll and wrapped it in JNI using the stubs from libshout-java
> (http://benow.ca/index?page=project&project=libshout-java). So, I have
> libshout basically working through Java on Windows XP.
>
>     I think everything is working ok, except shout_sync. This call
> returns immediately under the compilation setup above, instead of
> actually sleeping, screwing up the audio streams. This occurs in exactly
> the same way when I compile the ezstream utility off of the libraries as
> when I'm running libshout through Java, so I don't think it's
an
> artifact of the JNI process.
>
>     Looking at timing_sleep, the only relevant call there is select(),
> so I'm guessing the problem is in the invocation there, but I can't
> figure out what would fix it -- I've been programming Java too long,
I'm
> just glad I made it this far :)
>
>     I modified the JNI wrappers to expose shout_delay and have been
> handling sleeping on my own in Java, and everything seems to work
> perfectly. So this isn't a debilitating issue for me, but I'm
hoping
> someone on the list will recognize what the fix would be and can get
> shout_sync working at some point under MinGW. Thanks!
>
> -alan
>
> PS. under regular cygwin this problem doesn't occur; I need to go ming
> for JNI.
>
> _______________________________________________
> Icecast-dev mailing list
> Icecast-dev@xiph.org
> http://lists.xiph.org/mailman/listinfo/icecast-dev