[FFmpeg-devel] Small modifcation to libavformat/dvbsubdec.c

JULIAN GARDNER joolzg at btinternet.com
Wed Sep 18 00:09:33 CEST 2013

> From: compn <tempn at twmi.rr.com>
>To: ffmpeg-devel at ffmpeg.org 
>Sent: Tuesday, 17 September 2013, 17:06
>Subject: Re: [FFmpeg-devel] Small modifcation to libavformat/dvbsubdec.c
>On Tue, 17 Sep 2013 12:01:07 +0200, Reimar Döffinger wrote:
>>I hope I am not on the wrong track again with my current theory...
>i think his patch just conforms more to the spec.
>why not apply it , and wait to see if it breaks anything? :)
>i dont think anyone is insulting in this thread, just some
>frustrations. :)
>for michael and reimar : there is no sample that this patch fixes.
>for j00lz: michael and reimar like having samples for patches or knowing
>if this patch breaks any existing samples. i assume you've tested this
>change on multiple dvb samples and hardware?

I have tried on everything i can, but cant test a real world stream as the encode to dvb fails with the timestamp interleave problem, this was one thing i was going to work on.

>thats just how ffmpeg development has gone in the past. someone writes
>a demuxer/decoder. then a sample comes along that is broken. then a
>patch to fix it. repeat repeat repeat. but when there is no sample, its
>scary to us to change the behavior that has worked for 10 years.

But catering for BROKEN streams should not be the default, there should be a switch to say to the drivers to allow non-spec behaviour.

>we value your contributions j00lz. we are just scared of changes
>without samples.

Well i cannot give you a sample as the only ones I have are from broken encoders and under NDA as the companies dont want people to know about these problems.


More information about the ffmpeg-devel mailing list