Re: Pending 1.2 stuff

[Havoc Pennington <hp redhat com>]

"Matthias Clasen" <Matthias Clasen poet de> writes:
> > So I think you may be right; what we were originally saying is that
> > _FLOATING would be "the kind of window xmms and gkrellm are," and in
> > that case we could just leave it up to the WM to establish how such a
> > window relates to dock windows and other kinds of window.
> 
> So does this point to the need for a separate semantic type for
> "the kind of window xmms and gkrellm are" ? If so, how would that type
> be described ?

See the whole previous thread on this - there was some discussion of
state vs. type, and what this type might be called.

Personally I don't think the kind of window xmms and gkrellm are
should exist at all - I don't know about other WMs but these two apps
are the #1 source of bug reports for metacity, because they just do
not fit into the framework of "how the WM works" - I think they should
be docks/panels, or have a freaking titlebar.

Still given that they exist I'd like to have some sane and documented
way to deal with them.

> > There is one other possible use of "on top" I've seen, which is for
> > dialogs that you might want to keep on top for some reason; one case
> > of this is say the ssh-add password dialog on login, maybe you want to
> > keep that above other apps that start up on login.
> 
> I think this could be handled by setting the urgency hint and have the wm 
> keed urgent dialogs above normal windows at startup.

That does have the plus of keeping things semantic.

Havoc