[Unichrome-devel] New libXvMC wrapper.

[Thomas Hellström]

Alan Cox wrote:

>On Sad, 2004-09-11 at 08:24, Thomas Hellstrom wrote:
>  
>
>>The problem to be solved (for future compatibility) is the following:
>>1. Is the X server local? If not, best for now is to shut down.
>>    
>>
>
>Surely if not send bitstream over the protocol
>
>  
>
This would have been nice, since it would allow decoding to take place 
in the local X server of a
cheap EPIA board. Unfortunately, the protocol implementation in the X 
server is not complete, and just implements surface and context creation 
/ deletion and capability querying.

>>2. What is the name of the XvMC Driver? One option would be to load in 
>>turn a number of known drivers and try to create a context. This will 
>>most likely fail if the wrong driver is used.
>>    
>>
>
>This sounds more and more like XvMC and DRI should be talking and using
>the same kind of logic. 
>
>  
>
Yes, this would be possible by letting the X server side of 
XvMCCreateContext return the name
of the driver as part of the non-device-private part of the info record. 
This would require a slight
redesign of the protocol, and all existing drivers would need a change. 
At the very least this should be part of a new protocol revision, but 
for now I think we are stuck with the wrapper.

Another possibility would be, of course, to parse the Xf86Config file.


/Thomas



>_______________________________________________
>xorg mailing list
>xorg at freedesktop.org
>http://freedesktop.org/mailman/listinfo/xorg
>  
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.x.org/archives/xorg/attachments/20040912/9758ae10/attachment.html>