[Ffmpeg-devel] time for a release?

[Diego Biurrun]

On Sun, Jan 22, 2006 at 07:55:07PM +0100, Diego Biurrun wrote:
> > 
> > Where is that PSP patch?  Has anybody looked at it yet?
> 
> ... ping ...

I don't remember offhand whether this patch has been applied or not..

> Are there any other important patches which would be nice to get merged
> before the release?

The H.264 alignment thing from David Miller is still pending.  IIRC
Michael approved it modulo some testing.  Would be nice if somebody
could adopt, test and apply this patch.

The same goes for the other patches that Michael approved.  Let's try to
distribute the load a bit better.

> There are a few pending bug reports as well, I'm aware of the following
> ones (more or less in order of importance):
> 
>   Date: Wed, 18 Jan 2006 21:57:57 +0100
>   Subject: [Ffmpeg-devel] [BUG] H.264 crash
> 
> Loren, Michael?  Another crashing sample was posted today.

This one is still outstanding and rather severe.  A mail with a dirty
fix or possible workaround arrived yesterday.

>   Date: Mon, 02 Jan 2006 13:38:30 +0100  
>   Subject: [Ffmpeg-devel] [BUG] MPEG-4 file with artifacts
> 
> Michael?

Settled, the sample was broken.

>   Date: Mon, 02 Jan 2006 14:50:46 +0100
>   Subject: [Ffmpeg-devel] [BUG] TSCC wrong colors on PPC
> 
> Mike said he would try to fix this soon once he gets a bit more settled
> into his new home.

Fixed.

>   Date: Mon, 02 Jan 2006 15:04:22 +0100  
>   Subject: [Ffmpeg-devel] [BUG] MPEG-4 artifacts, possibly demuxer-related
> 
> Anyone?

Also settled, fixing this requires big restructuring or performance
loss.  It will have to remain unfixed for the moment.

> Are there important bug reports that I am missing?  Otherwise I would
> suggest preparing the release once we have these sorted out.

One thing that I have overlooked are the regression tests.  The ffserver
regression tests fail currently even on x86, on PPC regression tests are
completely broken, dunno about other platforms.  Putting out a release
that does not pass its own regression tests would be quite embarassing,
so this should IMO be addressed.  At least x86 should work.

Anything that I have missed?  Otherwise I think we're really getting
closer...

Diego