[PATCH xserver 2/2] randr: Use RRCrtcGetScanoutSize() in ProcRRSetScreenSize()

[Keith Packard]

Michel Dänzer <michel at daenzer.net> writes:

> Sounds like we only want patch 1 then?

Well, there's a difference between what the software can do and what the
protocol should require. The original reason for the RandR restriction
requiring a frame buffer large enough for all CRTCs was simple hardware
requirements - you can't scanout pixels you haven't allocated.

With shadow buffers, that restriction is no longer necessary. And, in
fact, it's obviously not what you want either -- when rotated at a non
right-angle, you generally want the whole image displayed in through the
CRTC with blank pixels filling in the corners.

Hrm. On consideration of the potential for minor arithmetic rounding
errors between client and server, I actually do think it would be best
to not have this requirement for arbitrary transformations. Otherwise,
the client would have to carefully replicate the precise operations used
in the server to compute the bounds. Given that the matrix provided is
fixed point (which causes visible problems for projective transforms),
the opportunity for error is even higher.

-- 
-keith
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <https://lists.x.org/archives/xorg-devel/attachments/20180110/65fe261b/attachment.sig>