[FFmpeg-cvslog] r11625 - trunk/libavformat/mov.c

Baptiste Coudurier baptiste.coudurier
Sun Jan 27 17:09:09 CET 2008

Reimar D?ffinger wrote:
> Hello,
> On Sun, Jan 27, 2008 at 04:15:01PM +0100, Baptiste Coudurier wrote:
>> Reimar D?ffinger wrote:
>>>> This is awfully hackish. IMHO codec id CODEC_ID_MOV_TEXT must be
>>>> implemented and this mess must be done at the codec level.
>>>> Please revert !
>>> A CODEC_ID_MOV_TEXT is rather pointless with nobody supporting it, and
>>> given the format I have some doubts it will ever be supported.
>> This is a bad justification for hacks.
> It seems to be a good enough justification for all the hacks being
> applied to the extradata...

Humm, to parse atoms containing extradata ? Well having multiple atoms
is messy yes, though there is 'glbl' now ;)

>>> In addition this was the way thing were handled for Matroska subtitles
>>> at first as well...
>> One demuxer already doing hacks is not a justification either.
> No, but the fact that it was considered an ok way back then did not make
> me consider this a hack.
>>> And lastly, while I miss the specifications to say for sure, the
>>> subtitle format seems to be created on purpose the way that there is
>>> first a pure plaintext version and then a formatted text version in one
>>> packet, which makes this way to handle it seem much less hackish.
>> Data might be there on purpose and must be kept with stream copy.
> I can not deny that (though this as well does not seem to apply to
> e.g. extradata?)

I don't see what you mean here. If you mean exporting data at the end of
stsd, that's what mov_read_extradata does, and handles multiple atoms in
extradata. The mov_read_glbl exports extradata for codecs not supporting
atoms, to avoid breaking decoding.

Still hacks are hacks and should be removed, this is not a justification
to add more.

>> Extracting data can be done at bitstream parser and better at codec
>> level if you want to get AVSubtitle structure.
> I'd have to investigate to even know of bitstream filter can be used
> properly (and more importantly how).

I even see that we don't have a CODEC_ID_TEXT subtitle decoder in lavc.
It should use AVSubtitle IMHO.

>> Please revert.
> May I at least make it a #if 0 to keep as documentation of the format?

You could point the url to quicktime specifications if someone wants to
properly implements it.

Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312

More information about the ffmpeg-cvslog mailing list