[FFmpeg-devel] [PATCH] play aiff without common tag

Reimar Döffinger Reimar.Doeffinger
Mon Jun 25 20:05:57 CEST 2007


Hello,
On Mon, Jun 25, 2007 at 06:54:15PM +0100, M?ns Rullg?rd wrote:
> Michael Niedermayer <michaelni at gmx.at> writes:
> > On Mon, Jun 25, 2007 at 07:21:20PM +0200, Michael Niedermayer wrote:
> >> On Mon, Jun 25, 2007 at 05:51:59PM +0200, Reimar D?ffinger wrote:
> >> > attached patch fixes
> >> > http://samples.mplayerhq.hu/AIFF/invalid_nocommon.aiff.
> >> > While according to the specification says this file invalid I think it
> >> > is preferable to support it anyway, esp. because leaving values
> >> > uninitialized and even causing a division by zero is really bad anyway
> >> > (and it even leaves e.g. MPlayer thinking this is a video file...).
> >> 
> >> looks ok
> >
> > hmm ive just noticed baptiste is maintainer of aiff ...
> > so its his decission
> >
> > but id like to repeat that ffmpeg CANNOT reject files just because they
> > are not conforming to specs, we would be rejecting >50% of all files
> > otherwise and ffmpeg would be unuseable
> 
> For uncommon types of nonconformance we should fail with an error
> unless -strict -1 was specified.
> 
> More common errors we can let pass with just a warning at strictness
> level 0, but no higher.  This is only to reduce the support burden
> for people too lazy to read and follow instructions.  Ideally we
> should not tolerate any deviation from spec by default.

Then IMO make strict = 1 the default, but according to the MPlayer
description of strict (is there one in ffmpeg?) strict = 0 only means that
libavcodec specific stuff is disabled but nothing more, which I'd
interpret as "do as everyone else".
And I am not sure that adding code to print warnings about every badness
is such a good idea, unless we can make adding those simpler.
Just "aborting" based on strict is easier.

Greetings,
Reimar D?ffinger




More information about the ffmpeg-devel mailing list