Re: [jirka 5z com: Re: KDE 2.0 impressions]

[Owen Taylor <otaylor redhat com>]

Owen Taylor <otaylor redhat com> writes:

> Havoc Pennington <hp redhat com> writes:
> 
> > Mary Dwyer <Mary Dwyer ireland sun com> writes:       
> > >      I don't think I understand what you're saying Havoc. Even if the            
> > >      DESKTOP_LAUNCH_ID and X property are defined and set as you say, 
> > >      wouldn't you still be using the LD_PRELOAD hack to end the feedback display
> > >      for that client?
> > >
> > 
> > I'm saying that we should have a well-defined protocol, so that GTK
> > for example would set the X property on its windows itself. Then GTK
> > apps wouldn't need the LD_PRELOAD. I would expect Qt to also follow
> > the protocol. Then the LD_PRELOAD hack will just be for legacy apps.
> 
> Can't we simply use the _NET_WM_PID property defined in the new
> Window manager protocol? I'm willing to add support for this
> in GTK+-1.2.9 if necessary.

OK, in discussing this with Havoc, he points out that using the
PID doesn't  with wrapper scripts such as the ones typically
used for Netscape.

However, it does appear it is the mechanism used by KDE:
 
 http://mail.gnome.org/archives/wm-spec-list/2000-September/msg00000.html


So, perhaps we need something more, like setting an environment
variable. But the requirements, I think are:

 - Simple; no use of CORBA, etc. 

 - The part in the Panel should to interoperate with the KDE-2.0
   solution, even if we have something better.
 
 - If the properties we use are in the _NET namespace, we must
   get a consensus on wm-spec-list for addition of the properties
   to the next version of the wm-spec before we do it.

 - Any LD_PRELOAD hacks must be explicitely enabled in the desktop
   entry. Indescriminately LD_PRELOAD'ing everyhwere is to
   be avoided.


If we can come up with something, and come up fairly with it fairly
quickly, I'll add the necessary code into GTK+-1.2.9 so that
all GTK+ and GNOME applications work with it out of the box.

Further discussion of this should probably occur on wm-spec-list.

(http://mail.gnome.org/mailman/listinfo/wm-spec-list)

Regards,
                                        Owen