Re: Sticky vs All Desktops

[Havoc Pennington <hp redhat com>]

On Thu, Jan 23, 2003 at 07:49:44PM -0600, Ben Jansens wrote: 
> I also have a further question on All-Desktops support. The spec
> itself does not mention that desktop windows or dock windows should me
> automatically made omnipresent.
> 
> In the case of the gnome-panel (dock type), it sets itself on all
> desktops. Whereas in the case of nautilus, it just sets the desktop
> type, and expects to be made omnipresent by the window
> manager. Perhaps some clarification in the spec could be added as to
> which types expect (SHOULD?) be made omnipresent, and which should
> not. Or perhaps none should and nautilus should be required to set
> itself to all desktops.
> 
> I like having each application decide to be omnipresent or not because
> this would let you have a 'desktop' type window that was not
> omnipresent without it having to change desktops after
> mapping. However, this also might break any netwm desktop apps out
> there (heh, is there any besides nautilus and kdesktop yet?).

You're right this should be clarified a bit. In general we *do* want
the WM to decide a lot of things based on semantic type - that's the
point of having the semantic types, so WMs can decide on the exact
behavior instead of hardcoding it in apps.

However, sure I can see that it might be useful for a desktop window
to be limited to a single space, for example for implementing
"different background for each desktop"

Havoc