Re: Decorations (again)

[Paul Warren <pdw ferret lmh ox ac uk>]

On Mon, 8 Nov 1999, Marko Macek wrote:

> > I think not.  We should definitely take this opportunity to get all WM
> > related hints in one, consistent, place.
> 
> I agree. I suggest we also define a packing for other X hints (wm should
> still use fallback properties if hints are not in _NET_HINTS):
> 
> WM_PROTOCOLS, WM_CLIENT_LEADER, WM_TRANSIENT_FOR, WM_HINTS,
> WM_NORMAL_HINTS

Agreed.

> > > Another topic are modal dialogs. Can't remember whether we discussed that. I
> > > was wondering whether a hint MODAL_FOR (similar to transient_for) may be useful.
> > > It would give the WM a hint which window to activate (there's no point in
> > > activating a window that has a modal_for window open).
> > 
> > Sounds reasonable.  Is this best implemented as a MODAL_FOR hint, or
> > using the existing TRANSIENT_FOR mechanism, and simply setting a MODAL
> > flag on the window?
> 
> Single flag (we need OWNER_MODAL, GROUP_MODAL and maybe DESKTOP_MODAL
> flags, or perhaps a new subhint _NET_WIN_MODALITY (3 atoms above)).

Sounds good.

> This also implies a WM_TRANSIENT_FOR=Root/None extension. Most wms
> already have handle the case of =Root gracefully it because of some
> "broken" programs set it to various strange values (root, self, none,
> ...)

I think the consensus is to put the WM_TRANSIENT_FOR=Root extension into
the spec, despite the minor ICCCM infringement - it's what everyone does
anyway, why not make it official :-)

Paul