Re: Window History Placement

[Havoc Pennington <hp redhat com>]

On Fri, Jan 10, 2003 at 04:02:36PM +0100, Matthias Clasen wrote:
> > 
> >  Looks like I misunderstood one part of KWin code. But still, windows can 
> > explicitly request a position, which should disable WM's placement policy,
> > 
> > no?
> 
> I think so, but I guess Havoc has a different opinion.

What opinion are you saying I have? ;-)

If we're talking about USPosition/PPosition, my experience is:

 - you basically have to always honor USPosition/PPosition or some
   apps break, some of them are ridiculous - such as Java needing 
   PPosition honored because it uses managed windows for popup menus -
   but sometimes PPosition is legitimate.

 - so when an app specifies PPosition or USPosition, metacity applies
   its hard constraints (can't move the titlebar offscreen for
   example), but otherwise leaves the position as requested

 - if there's no PPosition/USPosition, metacity uses a placement 
   algorithm

With "window history" I would expect that the window history is
basically part of the placement algorithm - instead of "first fit" we
"restore previous position" - so would only be used when we were doing
placement. (For the position aspect - you might restore say maximized
state in all cases.)

Havoc