composite manager thoughts

[Keith Packard]

Around 19 o'clock on Dec 28, Allen Akin wrote:

> Is the intent that the compositing manager never perform geometric
> transformations when displaying the contents of application drawing surfaces?

Of course not; we've got lots of potential room to play when we start 
allowing this kind of manipulation.  The major impediment today is the 
lack of an input model -- we can transform the output, but have no way of 
mapping global mouse coordinates to intra-window coordinates.

> For example, scaling down a window containing text might make the text
> unpleasant to read unless the app re-renders with appropriate hinting.  If
> the scale factor is nonuniform (e.g. a perspective transformation or an
> application of a window to a curved surface), even the specification of the
> necessary parameters is challenging.

While we want to permit this kind of interaction at some point, I think 
it's important now to focus on the top level interaction semantics and 
ensure that existing applications continue to work; figuring out how the 
compositing manager + X server + window manager work together seems like a 
big enough puzzle without also dealing with how to get applications 
rendering themselves accurately across a spherical surface.

At first glance, I think it will be pretty hard for us to accidentally
break this, but we should remain cautious as we do the upper level design.

-keith