[FFmpeg-devel] [PATCH] ALSA for libavdevice

Stefano Sabatini stefano.sabatini-lala
Thu Dec 25 11:29:16 CET 2008


On date Sunday 2008-12-21 21:12:35 +0100, Michael Niedermayer encoded:
> On Sun, Dec 21, 2008 at 08:10:37PM +0100, Luca wrote:
> > Hi,
> > 
> > sorry for the delayed response
> > 
> > On Fri, 2008-12-19 at 10:59 +0100, Nicolas George wrote:
> > > L'octidi 28 frimaire, an CCXVII, Luca Abeni a ?crit :
> > > > So, it seems to me that the alsa library is still not stable enough for
> > > > what concerns the snd_pcm_htimestamp() function. :(
> > > 
> > > All I can say is that it works at home (Debian Testing).
> > 
> > Well, I tested various ubuntu installations, but none of them provides
> > snd_pcm_htimestamp(). I've been able to compile the new alsa code only
> > on debian unstable... I tested it on various different machines with
> > different audio cards, but snd_pcm_htimestamp() always blocks and ffmpeg
> > hangs.
> > 
> > 
> > >  Remotely testing ALSA programs is not easy.
> > 
> > I know... I suppose this problem depends on the ALSA library version,
> > but also on the driver (and on the hardware). 
> > 
> > 
> > > I guess that means we will have to fall back to the av_gettime-based
> > > timestamp, probably with a call to know the delay to ensure A-V sync.
> > 
> > Well, I think that a configure test for snd_pcm_htimestamp() (or for the
> > proper libasound2 version) is surely needed to avoid breaking
> > compilation.
> > A fallback solution based on gettime+compensation would be even
> > better...
> > 
> > 
> > > Shall I work on a filtering function and submit it as a prerequisite patch?
> > I do not know... I hope a alsa input with proper checking for the right
> > alsa version can be accepted first, but this is up to Michael.
> 
> if its possible to test if snd_pcm_htimestamp() is working in configure
> (that would require detecting that it hanged ...)
> alternatively if someone happens to know which version/driver/hw works ...
> if neither of these then i dont see how we could add snd_pcm_htimestamp()
> based input support

Maybe complaining upstream could help, if we get it fixed then we just
have to check for some ALSA version, rather than put random magik in
configure.

Regards.
-- 
FFmpeg = Fundamental and Forgiving Marvellous Peaceless Epic Gadget




More information about the ffmpeg-devel mailing list