[FFmpeg-devel] [RFC] LOAS (aka LATM) demuxer

[Kieran Kunhya]

--- On Mon, 2/8/10, Martin Storsj? <martin at martin.st> wrote:

> From: Martin Storsj? <martin at martin.st>
> Subject: Re: [FFmpeg-devel] [RFC] LOAS (aka LATM) demuxer
> To: "FFmpeg development discussions and patches" <ffmpeg-devel at mplayerhq.hu>
> Date: Monday, 2 August, 2010, 14:17
> On Sun, 1 Aug 2010, Kieran Kunhya
> wrote:
> 
> > > It is on its own not very useful since I doubt
> loas
> > > files/streams exists
> > > besides the ones I've extracted out of mpeg-ts
> samples. It
> > > needs to be
> > > integrated into the mpeg-ts demuxer. I'm not
> convinced that
> > > demuxer
> > > chaining is the best way to do it.
> > 
> > Without chained demuxing how would you handle LATM in
> rtsp which has 
> > been seen in the wild (youtube iirc)? Presumably
> someone will be crazy 
> > enough to use LATM in mp4 in the future too.
> > 
> > I'm not saying these formats are in any way sane but
> there's going to be 
> > a ton of code duplication otherwise. 
> 
> I guess it varies from case to case what is the most sane
> way to use it. 
> We already use chained demuxers for some RTP formats, so I
> don't think 
> it'd be a problem to use it for this format, too. But for
> mpegts, I can 
> see how it feels a bit complicated.
> 

SMPTE 337M and 302M are worse in my opinion so I don't mind a bit of discussion to get the best idea out there. I think Michael's idea is workable but it adds a layer of complexity that imo is unneeded. It might be the only decent way to do it without modifying a lot of the API or lavf/lavc.