[Bug 735379] mpeg2dec: use stride information from GstVideoFrame

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=735379

--- Comment #11 from Nicolas Dufresne (stormer) <nicolas.dufresne at collabora.co.uk> ---
(In reply to Sebastian Dröge (slomo) from comment #10)
> You should still check if it doesn't change for future buffers, and error
> out if it does. Also the stride might change after a reconfiguration (e.g.
> because now we got a different pool from downstream).

I still think a pool shouldn't change it's alignment dynamically. Of course, in
the following patch I'll have robust code. I'm not done yet as you can see, as
I just re-oriented the code toward a design that I know is well covered, and
tested.

> 
> Why did you remove the crop support?

The implemented crashed, and was wrong in many ways. There is nothing to reuse
there really. I'll re-add later, though I'd focus on theora. Unlike
video-alignment method, CROP meta is only rarely supported. Nearly all the
transform element drop that from the query unless passthrough, even videocrop
does not yet implement it. When we get the in theoradec (since there is no
other solution there, unless the crop is chroma aligned). I'd be happy to
re-add this feature later. Meanwhile, video-alignment is a much more powerful
optimization, and I don't believe we should care to much about CROP here.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.