XComposite input redirection/transformation proposal

[Keith Packard]

On Tue, 2008-02-19 at 16:26 +0100, Dennis Kasprzyk wrote:

> > I fear it's not possible. Compositing managers allow windows to be
> > displayed multiple times (e.g. once in the pager, once normally) and
> > some allow even more weird multiple instances of the same window (take a
> > look at Metisse). I'm sure one day they'll want to enable input on some
> > or all of these instances, meaning bijection isn't an option.

Perhaps not a true bijective mapping, but some kind of complete mapping
in both directions is required by the protocol. In the multiple
presentation case, you might simply have to pick one of the
representations for the child->parent mapping. In the pager case, we've
got something more complicated going on as (potentially) the child is
presented in more than one parent window. I don't know how to make that
work...

> In a previous mail to this thread I've proposed a way how to get "cloned"
> windows (InputOnly redirection Windows) and the bijection. But the system
> would be more complicated.

This might also make the multi-parent case tractable, as described
above.

In any case, the first thing I'd like to see is a design which satisfies
all of the core protocol requirements for children represented precisely
once within their natural parent. Once we have that, we can start
exploring how to provide for multiple representations, and children
represented within other windows.

-- 
keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://lists.x.org/archives/xorg/attachments/20080219/28b23a14/attachment.pgp>