[Bug 764316] vaapidecode: simplify downstream negotiation

GStreamer (GNOME Bugzilla) bugzilla at gnome.org
Wed Mar 30 09:17:43 UTC 2016


https://bugzilla.gnome.org/show_bug.cgi?id=764316

--- Comment #5 from Víctor Manuel Jáquez Leal <vjaquez at igalia.com> ---
(In reply to sreerenj from comment #4)
> Hm, these patches are causing more regressions . Not only the vp9 sample i
> have shared before.

Can you point me to a failing test? Or email me a vaapidecode:7 log?

> 
> Patch1 is trivial , just removing the function declaration, which is okay to
> push 
> if you want.

Yes.

> But others  require careful review and regression testing. So please don't
> push those now.
> we can think of removing more global state variables, but on the other hand
> I have a feeling that it is more readable when keeping separate VideoInfo
> for decoded_surface and display_info. Better optimizations also possible by
> removing Caps variables from GstVaapiDecode structure and utilize the
> GstVideoInfo instead.

The only argument of keeping a output_{width,height,format} variables instead
of GstVideoInfo is because of memory saving. But I can agree on change it.

What I can't understand is the reason of having both decoded_info and
display_info, when we only care about the downstream caps negotiation.

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


More information about the gstreamer-bugs mailing list