[FFmpeg-devel] [PATCH] audio conversion clipping/overflows

Michael Niedermayer michaelni
Tue Mar 16 11:55:01 CET 2010


On Tue, Mar 16, 2010 at 07:24:34AM +0100, Reimar D?ffinger wrote:
> On Mon, Mar 15, 2010 at 06:31:21PM -0400, Ronald S. Bultje wrote:
> > On Tue, Mar 2, 2010 at 7:36 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > > also if you apply this, make sure you remove the cliping from every decoder
> > > that matches SAMPLE_FMT_FLT
> > 
> > OK, see attached patch. Maintainers of relevant decoders (Colin,
> > Benjamin, Vitor, Kostya, Sascha, Roberto, Reynaldo), please bug me if
> > there's a very specific reason that you would like me to NOT remove
> > clipping from your code, else I'll apply this in +/- 3 days.
> 
> Uh, isn't that kind of an  API change for everyone not using the internal
> format conversion?
> I suspect it might break a lot of programs that assume the clipping is done...

Lets see
someone writes a program and realizes that our decoder does not output samples
outside -1..1 so he designs his program to fail for samples outside.
This is called idiot not API change
next someone will design a program that has a buffer oveflow if the longest
codec names gets longer ...

nowhere do we claim output is cliped. Its undocumented behavior, undocumented
behavior is never part of a guranteed API/ABI that would be totally
unpractical as nothing could ever be changed
If someone wants to depend on undocumented behavior there is ffmpeg-dev where
one can ask if its just pure luck its like it is or if we gurantee the
specific bahavior

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

In fact, the RIAA has been known to suggest that students drop out
of college or go to community college in order to be able to afford
settlements. -- The RIAA
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100316/9478e67c/attachment.pgp>



More information about the ffmpeg-devel mailing list