[gstreamer-bugs] [Bug 578278] gst-ffmpeg: assign offsets (from upstream) to outgoing buffers

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=578278
  GStreamer | gst-ffmpeg | git

Mark Nauwelaerts <manauw> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |manauw at skynet.be

--- Comment #12 from Mark Nauwelaerts <manauw at skynet.be> 2010-01-05 17:28:55 UTC ---
(In reply to comment #5)
> commit 73d227766a1be096253ebbdfd7fc4df5ddfe8f7b
> Author: Руслан Ижбулатов <lrn1986 at gmail.com>
> Date:   Sat Jul 18 18:53:22 2009 +0400
> 
>     Codec frame delay fix and trailing zero-length frame fix
> 
>     Takes codec frame delay into account (roughly the same way it does for
> times
>     A special hack to allow trailing frame with timestamp=segment.stop to be
> dis
> 
>     Fixes bug #578278.

After a number of versions of this bug/patch, the above 'hack' still seems to
be in there.  But that 'hack' actually breaks/overrides normal segment
semantics, and would have some nonlinear-filtering pipeline (e.g. entrans,
gnonlin) end up with different (more) output than it would before.  Similarly,
video stream may end up with more frames than there is corresponding audio data
properly clipped to the same segment.

[the hack considers a buffers starting at segment->stop to be part of the
segment, and fiddles with such buffers' start and duration, etc.
Normal segment clipping does not consider it to be part of segment]

So, unless there is some very good reason to have 'special hacks' breaking
standard behaviour, I feel like reverting that part.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.