Re: pseudo transparency

[Olivier Chapuis <olivier chapuis free fr>]

On Fri, Jan 11, 2002 at 06:42:12PM -0500, Owen Taylor wrote:
> 
> 
> I think the ESETROOT method is actually pretty workable, if a little
> hackish.  I'd be willing to write it up for freedesktop.org. (Havoc
> says he doesn't want it in the WM spec. I tend to agree, if for no
> other reason than it really isn't related to the window manager.)
>

If you write a document, IMHO, it will be good to document the
_XSETROOT_ID(Pixmap) property too.
Some set root programs as xli use it. The rules are simple
(these are just result of some observations):

1. If this property is set the corresponding pixmap should be
"destroyed" (with XKillClient) before changing the root background.
xv does this but (unfortunately) it does not set this property.

2. After setting the background the _XSETROOT_ID should be set
to indicate to the interested programs that the root background
has changed. There is no specification on the pixmap on which
_XSETROOT_ID point on (it may be None). With xli _XSETROOT_ID is
a depth 1 1x1 pixmap (with other programs _XSETROOT_ID is a
pixmap smaller than the screens with which the root background
has been tiled). 

3. Set root programs which use this method do not set
the _XROOTPMAP_ID property. So poor programs :o) that use
parental relativity may monitor both _XROOTPMAP_ID and
_XSETROOT_ID for background change without having to
redraw itself 2 times at each background change.

This method is useful for environment/users which want to
use the parental relativity method only to achieve pseudo-
transparency (yes, this save some memory).

If you want I can try to write down something more clear
and in a better English in the continuation of your document.

Regards, Olivier