[PATCH] Fixes v5: Cursor barriers

[Adam Jackson]

On Tue, 2010-12-07 at 11:26 +1000, Peter Hutterer wrote:

> > +	Servers supporting the X Input Extension version 2.0 or higher
> > +	respect these barriers for any cursor on relative motion events.
> > +	Absolute positioning devices do not obey these barriers as there's
> > +	no benefit to target acquisition to do so.
> 
> Is it worth making it a per-device barrier with the usual XIAllDevices,
> XIAllMasterDevices. Though it could be interesting handling a blocked master
> device on a non-blocked SD...

I can just about imagine a use for being able to make barriers
per-master, it'd give you a cute way of confining people to different
outputs.  Slaves don't seem that interesting (see Daniel's comment).

Maybe just leave an XID slot in the request and note that values other
than None/XIAllDevices may mean something else in a later Fixes version?

> > +DestroyCursorBarrier
> > +
> > +		barrier:		    BARRIER
> > +
> > +	Destroys the named barrier.
> > +
> > +	Errors: Barrier 
> 
> Do we need a new error here? BadValue would be sufficient, especially given
> that the request has no other value.

Traditionally, a new type means a new error.  Keith?

- ajax