On Mon, Feb 11, 2008 at 07:14:11AM -0600, Rich Coe wrote:
> On Tue, 5 Feb 2008 13:33:37 +0100
> Dominik Vogt <[EMAIL PROTECTED]> wrote:
> > On Tue, Feb 05, 2008 at 05:50:00AM -0600, Rich Coe wrote:
> > > This patch fixes a problem with recently un-mapped windows.
> > 
> > You can't do it this way.  There is no guarantee that the stored
> > window ids refer to any known window; X may recycle window ids at
> > any time.
> 
>     I'm not quite sure what you're objecting to.  As windows are created
>     and mapped, the window ids get added to FvwmContext.  When the windows
>     are unmapped, the window id gets moved from FvwmContext to LruFvwmContext.
>     When the window is mapped again, it gets moved from LruFvwmContext back
>     to FvwmContext.

When a window is unmapped, the application usually destroys it,
and fvwm can not know whether that has happened.  If later a
window with the same window id is mapped, it may be the same
window, or it *may* be a different window that was created later
by a different application and that happened to get the same
window id as the earlier window that was destroyed.

> > Can you give me example instructions so that I can play with the
> > problem myself?  In what context do you get the crosshairs?
> 
>     If you pass in a "0" window id to a command, fvwm prompts the user
>     with a cross-hairs to select the window.  The problem is that the 
>     fvwm code is treating 'not an active mapped window id' the same as
>     "0" window id.  
> 
>     I'll make a test case that shows this behaviour.  I've just been
>     distracted on another issue.

Yes, please.

Ciao

Dominik ^_^  ^_^

-- 
Dominik Vogt

Reply via email to