[Mplayer-cvslog] CVS: main/libmpdemux asfheader.c,1.28,1.29

[Alex Beregszaszi]

On Tue, Feb 05, 2002 at 07:06:01PM +0200, Arpi wrote:
> Hi,

Also this patch should be reversed (at least the select the biggest bitrate stream
part)

> 
> > this biggest-bitrate-stream selection is a good idea, but it should be
> > optional! The 'preferred' stream is always with-the-lowest-id!
> disagree
> 
> teh current (i mean old) stream selection was the first-packet.

Yes, also in the most files first-packet = lowest-id, i meant the same as
you, but with bad definition.

> so teh first audio packet's id was selected as audio id and teh first video
> packet's id as video id. it workarounded broken files with many headers but
> only 2 streams.
> 
> anyway, wa sthis patch tested for this? good samples are the files generated
> by asfrecorder.
> 
> > I think all this stream id selection should be done commonly in the demuxer
> > code, all demuxers should set the available streams (with flags?) and then
> > we should select, optionally with this biggest-bitrate or lowest-id method.
> > (and other methods are welcomed, i think)
> not all files has header for streams

Also every demuxer uses new_audio/new_video.
We should use these headers.

> 
> > Note: also we should select lowest-id streams in mpg demuxer too, i've got a
> > DVD, with two audio streams, id=128 - movie audio, id=129 director's issue,
> > good, but id=129 comes first in the stream, se auto-selection selects it,
> > instead selecting the real id=128 stream.
> no way to find the lowest id unless demuxing a big enough part of the file
> _before_ teh actuall decoding...

Also my idea should be optional only.

> 
> 
> A'rpi / Astral & ESP-team
> 
> --
> "I don't RTFM? Wow. What's the meaning of this? It's new for me."
> 	-- Martin Baum, a tipical MPlayer user...
> _______________________________________________
> Mplayer-cvslog mailing list
> Mplayer-cvslog at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/mplayer-cvslog