"Extended Screen Info"-extension / was: Re: Vendor patches for X11R6.8.2...

Alan Coopersmith wrote:
> > Maybe not exactly what you wanted, but the patch in B1846 handles
> > a minor issues with xinerama on display walls and multi-input monitors.
> > We been using it internally under Xdmx, but it it'd be nice to
> > have it folded into trunk.
> >
> > On Nov 10, 2004, at 9:04 PM, Roland Mainz wrote:
> >
> >> Are there any patches out there used in the various X11 vendor releases
> >> which may be interesting for the X11R6.8.2 release (for example SuSE has
> >> commited lots of stuff to Xorg trunk which are also present in SuSE 9.2,
> >> some of these patches may be something nice-to-have for X11R6.8.2) ?
> >> If anyone has such patches floating around - please open a bugzilla bug
> >> (if there isn't anyone yet for that issue), attach patch, commit to
> >> trunk and then request approval for X11R6.8.2...
> 
> I'm not so comfortable with this patch, though perhaps I could be convinced.
> 
> First, for the 6.8.2 release, just how much do we intend to put in?  This
> seems to be crossing the line from important bug fix to new feature.
> 
> And for this specific functionality, is this really the best way to solve it?
> Shouldn't the applications querying for Xinerama be modified to not use their
> special behaviour when not wanted?   This seems to fool everyone to work
> around a few broken applications.   I don't know about Xorg, but I know for
> Xsun this would break apps that have other reasons for needing to know about
> what part of the screen maps to which underlying frame buffer (such as Sun-DGA,
> which is similar to Xorg's DRI, or Sun OpenGL).
> 
> Long term, we really need to decouple the long assumed 1-to-1 mapping of
> Xserver screen to physical screen.   Xinerama breaks it, but so do multi-head
> cards, multi-input monitors, multi-head projection systems, display walls,
> and various other things.   We were kicking around for a while inside Sun,
> but never got around to implementing, an extension that not only reported
> physical screen boundaries, but also other characteristics clients may need
> to know, like subpixel layouts of LCD's and similar devices (RGB vs. BGR,
> horizontal vs vertical, etc.).

What about doing something like an "Extended Screen Info" extension
(similar to the EVI (=Extended Visual Info) extension ? It may be a nice
thing to have since stuff like Xprint also changes screen properties
on-the-fly (actually per print job, document and/or even per-page) and
having a general API for that may help application developers a lot...

... so where should be move that discussion to - Xorg or Xorg_arch (I
can make a list of requirements from the Xprint side (like: Need an
event system for screen changes including "context" tokens in the event
(to define for which Xprint context the event is...)), Alan may do a
list for Xinerama - and someone has to look at RANDR) ? :)

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) [email protected]
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 7950090
 (;O/ \/ \O;)