On Tue, 02 Feb 2016 20:15:56 +0200
Rémi Denis-Courmont <[email protected]> wrote:

> Le 2016-02-02 17:26, wm4 a écrit :
> > On Tue, 02 Feb 2016 14:53:46 +0200
> > Rémi Denis-Courmont <[email protected]> wrote:
> >  
> >> Le 2016-02-02 14:29, wm4 a écrit :  
> >> > I'll ask my favorite question for vdpau related things: how does   
> >> this  
> >> > handle display preemption?  
> >>
> >> I think that there are no reasonable ways to handle display 
> >> preemption.
> >> As soon as libavcodec handles anything more than individual
> >> VdpDecoderRender calls, then display preemption handling becomes
> >> impractical, if at all possible. This is because there are no ways 
> >> to
> >> ensure that the application rather than libavcodec gets The One
> >> preemption error.
> >>
> >> Now, I don't mean to say that we should necessarily ignore display
> >> preemption. However, if we want to handle display preemption, I 
> >> think we
> >> have to create a common VDPAU wrapper shared by libavcodec and
> >> VDPAU-capable libavcodec applications. With that, we can (among 
> >> other
> >> things) catch the preemption error safely and, say, make all later 
> >> VDPAU
> >> calls fail safe until objects are destroyed.
> >>  
> >
> >  
> >> Conversely, if we stick to plain libvdpau as it currently exists, I
> >> think we have to forget about display preemption and pretend it does 
> >> not
> >> exist.  
> >
> > I can't forget that it exists, because it will be triggered by e.g.
> > display mode switching.  
> 
> All I'm saying is that either a VDPAU wrapper is involved, or 
> preemption is not handled. If you think that preemption must be handled, 
> then that means you need a VDPAU wrapper.
> 

If you write such a wrapper, I'll be all for using it. Otherwise, I'll
prefer a solution that is somewhat more light-weight and less intrusive.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to