[FFmpeg-devel] [PATCH 07/19 v2] avformat/hls: parse ID3 timestamps for elementary audio streams
Michael.Schenk at albistechnologies.com
Thu Apr 10 11:41:47 CEST 2014
first of all thanks for the effort on HLS for supporting alternative rendition! However
it looks like certain HLS servers do not like the seek which was added in the libavformat/hls.c
Example Anevia Encrypted URL: http://demo.anevia.com:3128/vod/disk1/content2/ss-hls-c-fixed/content2.m3u8
request: GET /vod/disk1/content2/ss-hls-c-fixed/content2-audio%3D96000-video%3D257000-1.ts HTTP/1.1
header='HTTP/1.0 200 OK'
header='Date: Thu, 10 Apr 2014 09:15:46 GMT'
header='Server: Apache/2.2.24 (Unix) mod_ssl/2.2.24 OpenSSL/0.9.8g DAV/2 PHP/5.3.24 IISMS/4.0 mod_wsgi/3.4 Python/2.7'
header='Last-Modified: Tue, 11 Mar 2014 22:19:21 GMT'
header='X-Cache: MISS from demo.anevia.com'
header='X-Cache-Lookup: MISS from demo.anevia.com:3128'
header='Via: 1.1 demo.anevia.com:3128 (squid/2.7.STABLE9)'
Unable to seek to offset 0 of HLS segment 'http://demo.anevia.com:3128/vod/disk1/content2/ss-hls-c-fixed/content2-audio%3D96000-video%3D257000-1.ts'
Failed to open segment of playlist 0
Error when loading first segment 'http://demo.anevia.com:3128/vod/disk1/content2/ss-hls-c-fixed/content2-audio%3D96000-video%3D257000-1.ts'
Because it is not needed I propose to remove it and this will solve the issue.
Please find attached a patch
Von: ffmpeg-devel-bounces at ffmpeg.org [mailto:ffmpeg-devel-bounces at ffmpeg.org] Im Auftrag von Anssi Hannula
Gesendet: Sonntag, 6. April 2014 16:58
An: ffmpeg-devel at ffmpeg.org
Betreff: Re: [FFmpeg-devel] [PATCH 07/19 v2] avformat/hls: parse ID3 timestamps for elementary audio streams
05.04.2014 15:20, Anssi Hannula kirjoitti:
> HLS provides MPEG TS timestamps via ID3 tags in the beginning of each
> segment of elementary audio streams.
> v2: fix issues with streams that have multiple ID3 tags
> Signed-off-by: Anssi Hannula <anssi.hannula at iki.fi>
> Hi all,
> and sorry for the delay in getting this patchset in.
> I plan to apply the patchset (minus subtitle parts) later today or
> tomorrow, unless any issues are found (by me or anyone here).
Done, now let's see how much broke.
> The only major change is that I've dropped the separate hls_id3_audio
> subdemuxer again due to the various timestamp issues and the ID3
> timestamp mangling is now done in the hls demuxer itself. So this is
> similar to what the first version of the patchset had, except that the
> issues with it have been fixed.
> The patch in this message is the major updated part of this patchset
> version. intercept_id3() contains most of the new code, other parts
> were mostly present in one of the previous versions of the patchset.
> For other parts I've applied the changes previously mentioned and/or
> The full patchset is at:
> and with extra commits (mostly subtitles) that will remain in TODO:
> What is missing from the subtitle parts is to see if they could be
> handled in a cleaner way (i.e. without the demuxer reopening hacks),
> and some additional testing. I don't likely have time to work on that
> for at least the next few weeks, though, but patches welcome of course.
ffmpeg-devel mailing list
ffmpeg-devel at ffmpeg.org
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 1225 bytes
More information about the ffmpeg-devel