2008/10/29 Philipp Kolmann <kolm...@zid.tuwien.ac.at>: > Viktor Griph wrote: >> >> On Tue, 21 Oct 2008, Viktor Griph wrote: >>>> >>>> Question is now, is it a problem on Skype's side (according to the >>>> comments, they draw a window and set the withdaw state) or is it some bug >>>> on >>>> your side? >>> >>> That particular diff is related to a fix for gtk-windows disappearing >>> when using some new GTK praxis. >>> (See http://www.mail-archive.com/fvwm-work...@fvwm.org/msg01359.html) >>> >>> It's hard to say exactly what Skype does since it's closed source. I'll >>> try to debug it some to see if I'm able to reproduce the issue using bare >>> X-calls, and see if those calls are right or not. I still belive that the >>> above diff is right in accordance with the ICCCM2 specification, but without >>> knoing what Skype does it's hard to say if it is fvwm or Skype that breaks >>> the rules. >> >> After som preliminary debuging, it seems as if Skype does a >> XWithdrawWindow() call, and then waits for some event to happen before it >> maps the window again. I'm not sure what it wat's for, but obviously the dif >> you posted prevents that event from being sent, since Skype doesn't try to >> remap the window. I'll do some more debugging trying with the aim to find >> out which event Skype waits for, and once I know that I should be able to >> say if it should be able to expect that event or not based on the sequence >> of oprations that has been going on. >> > > Sorry, if I do bother you. Just wanted to check back, if you had any success > in this issue. If needed maybe I can get some contact with skype, to help > you solve this issue.
Well, I finally had time to look into this issue at depth and it turned out to be QT that waited for the removal of the WM_STATE property of the window before mapping it again, after unmapinging it while skype was setting up the size hints, and the qt considered the geometry invalid. I've committed a fix to CVS. /Viktor
