Re: mandatory _NET_WM_PING

[Havoc Pennington <hp redhat com>]

Philipp Lohmann - Sun Germany <Philipp Lohmann Sun COM> writes: 
> I'd rather have an additional property in which the application can
> tell which version of the hints the application supports; apart from
> overloading the ping protocol this would have the advantage for WMs
> supporting applications written according to earlier versions of the
> spec.

Me too.

> > Beside it obvious usefulness for killing hung windows.
> 
> How obvious is this usefullness really ? Does an app hang just because
> its stuck opening a file on a slow NFS-server ? Why should a WM decide
> that an app is hung ? The user will do that anyway if he's bothered by
> it and kill it; with automated killing you inevitably kill
> applications which are indeed not hung; as e.g. the gnome session
> manager likes to do with OpenOffice.org which happens to start longer
> than the default settings of the session manager allow and therefore
> kills it outright..

(That SM behavior is borked.)

What metacity does with _NET_WM_PING is pop up a dialog saying "this
application is not responding, do you want to kill it?" (rather than
just killing it in a timeout)

Havoc