[FFmpeg-devel] [PATCH] avfilter: make it possible to return eof from graph when any sink first receives eof

Paul B Mahol onemda at gmail.com
Wed Feb 17 20:53:26 CET 2016


On 2/17/16, Nicolas George <george at nsup.org> wrote:
> Le nonidi 29 pluviose, an CCXXIV, Paul B Mahol a ecrit :
>> From f4d05fabeffd30ba83b85ca844f6d583f3794bd9 Mon Sep 17 00:00:00 2001
>> From: Paul B Mahol <onemda at gmail.com>
>> Date: Wed, 17 Feb 2016 12:42:37 +0100
>> Subject: [PATCH] avfilter: make possible to receive eof when shortest
>> sink
>>  ends
>
> This is broken. There is no guarantee that EOF will arrive on the shortest
> sink first. Even without really pathologic cases, there may be a lot of
> buffering in some codecs or filters, so you could have the actually
> shortest
> stream delayed enough so that a longest stream would end first.

I mean, patch is meant to fix never ending processing when using
sources with unlimited duration in complex filter graph.

So it should read, breaks processing when first eof happens.

There is open ticket for this bug.

> Also, this is a part of the code I am changing in depth to get rid of
> recursivity.

I think that is unrelated.


More information about the ffmpeg-devel mailing list