On Thu, 13 Oct 2011 11:12:19 +0100, Thomas wrote: > Right -- and that's client-specific which means if they try and call that > when the window is iconfied, it won't work unless all other children have > been mapped -- essentially the window it's expecting to set focus for is in > the iconic state, not the normal state. The window wasn't found, and in > such cases, a BadMatch is always generated.
Ah, yes, I gathered something like that was the case. >> Now, looking in Emacs' sources, it seems that Emacs don't call that >> itself, rather the window manager does. > No. Ok, so how does whatever that is calling X_SetInputFocus() (gtk, I guess) know to call it when I move the mouse over the icon? > FVWM is not the cause of this from what you're telling me here. Great, thanks for your kind help. > Emacs *has* to handle the case of a BadMatch in its own code, is the only > thing I can think of. If it's not doing that (and do not expect me to go > anywhere near the source of Emacs to look), it's already a bug. By perusing the Emacs and gtk code, it seems to me that gtk tries to handle it (by discarding those kinds of XErrorEvents), but while gtk2 apparantly succeeded in doing so, gtk3 fails. I can avoid the crash by explicitly discarding those XErrorEvents directly in Emacs, but I am not sure that is the correct solution: * http://article.gmane.org/gmane.emacs.devel/145052 > That's not to say there's something inherent in GTK3 which is also > triggering this, but don't expect me to check that either. :P I was merely asking if anyone here could imagine something in FVWM being the problem (as I guess FVWM is the most unconventional thing in my setup, and nobody else seems to have reported this crash), so I could rule that out - I wasn't trying to command you to check anything. I am sorry that I seem to have expressed myself that way; it wasn't my intention. > Oh -- ISTR your mentioning running 2.5.30 -- if this is due to Debian being > out of date, you've no excuse to not upgrade to 2.6.3, and try this all > again. I am not the maintainer of the fvwm package in Debian, sorry. > Note that I am only-ever interested in corefiles from FVWM. FVWM doesn't crash, so I don't see how I could produce a core file from FVWM. > Applications in the way you've described them, simply do not fall under > FVWM's remit for bug-fixing. > FVWM is not here to circumvent such bugs. I never tried to imply that; I am sorry that my feeble english apparantly made it seem so to you. Thanks again for your generous help. Best regards, Adam -- "Gav Adam Sjøgren Strik" a...@koldfront.dk
