[FFmpeg-devel] MOV Muxer fix to respect the standard

[Reimar Döffinger]

On Fri, May 22, 2009 at 05:59:38PM +0200, Larbi Joubala wrote:
> M?ns Rullg?rd a ?crit :
> > Larbi Joubala <larbi.joubala at resonate-mp4.com> writes:
> >
> >   
> >> Reimar D?ffinger a ?crit :
> >>     
> >>> On Fri, May 22, 2009 at 04:59:55PM +0200, Larbi Joubala wrote:
> >>>
> >>>       
> >>>> Index: movenc.c
> >>>> ===================================================================
> >>>> --- movenc.c	(revision 18893)
> >>>> +++ movenc.c	(working copy)
> >>>> @@ -1013,8 +1013,11 @@
> >>>>      put_be32(pb ,0); /* reserved */
> >>>>      put_be32(pb ,0); /* reserved */
> >>>>      put_be32(pb ,0); /* reserved */
> >>>> -    put_byte(pb, strlen(descr)); /* string counter */
> >>>> +    if (track && track->mode == MODE_MOV || !track)
> >>>> +        put_byte(pb, strlen(descr)); /* string counter */
> >>>>
> >>>>         
> >>> if (!track || track->mode == MODE_MOV)
> >>>
> >>> but anyway I doubt this should depend on some per-track property.
> >>>       
> >>>       
> >> You've got some reference about this?
> >> I didn't see it in the QT MOV standard or ISO 14496-12 standard.
> >>     
> >
> > Eh?  It's in the description of the 'hdlr' box/atom.  Why did you
> > suggest the change if you couldn't find it in the specs?
> >
> >   
> Of course, I read the specs before submitting these corrections but you 
> have mentionned that the fix I submitted should depend on some per-track 
> property and  I just wanna know what kind of property.

You check track->mode, to me that means that at least in theory
different tracks could have different modes, and thus the properties
for different tracks (in the same file!) would be encoded differently.
That makes that patch a somewhat strange way to do it in my view...