--===============1616506449== Content-Type: multipart/alternative; boundary="----=_Part_63560_15992956.1224330458158" ------=_Part_63560_15992956.1224330458158 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline Its been about two weeks for me that I have not been able to run koffice. Sorry I can't be more precise. Speaking for myself, the whole world thanks you for looking into this. -David On Sat, Oct 18, 2008 at 3:24 AM, Andreas Pakulat wrote: > On 18.10.08 02:13:12, Cyrille Berger wrote: > > On Friday 17 October 2008, David Boosalis wrote: > > > Rafael. > > > Maybe some one more intimate with kword can address this issue. I've > seen > > > this problem for sometime with koffice in SVN and that's why I peaked a > > > little under the covers to see what it is the matter. Eventually they > have > > > to address it, or maybe it's just my system and this is all a mute > point. > > Nope it's not just your system, well, at least I have heard other people > > having crash in kfilewidget at startup. > > > > As for the fancy things we do: > > Well, the code you've posted doesn't trigger a crash in a simple KDE4 app, > so there's something else that triggers the problem. I'm attaching the > sample app as its pretty small (app generated from kapptemplate). > > That said if I understood correctly I should just start kword and it > crashes? Well it does, backtrace attached. And I think its quite clear now > why it crashes: The kio-job in the KFileWidget constructor causes an update > of the GUI, hence showEvent is called. Unfortunately the KDirOperator isn't > created yet at this point, but please read on. > > I don't understand though why the nested event loop causes GUI events to be > sent. As far as I know it shouldn't. And as I've recently seen a similar > problem with KJob (the DeferredDelete event, that was posted to the > main-app-loop got executed during a nested event loop and then also during > the next run of the main event loop) we might be facing a Qt bug here. Did > this crash start to happen "recently", i.e. after the 4.4.2 update to > qt-copy? > > What I also don't understand is why I'm getting an "unknown protocol > kfiledialog" with the sample application, but not with koffice, so that > might be the reason why the sample app doesn't crash. > > Andreas > > -- > You will have long and healthy life. > > > >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to > unsubscribe << > > ------=_Part_63560_15992956.1224330458158 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline
Its been about two weeks for me that I have not been able to run koffice.   Sorry I can't be more precise.  Speaking for myself, the whole world thanks you for looking into this.

-David




 

On Sat, Oct 18, 2008 at 3:24 AM, Andreas Pakulat <apaku@gmx.de> wrote:
On 18.10.08 02:13:12, Cyrille Berger wrote:
> On Friday 17 October 2008, David Boosalis wrote:
> > Rafael.
> > Maybe some one more intimate with kword can address this issue.  I've seen
> > this problem for sometime with koffice in SVN and that's why I peaked a
> > little under the covers to see what it is the matter.  Eventually they have
> > to address it, or maybe it's just my system and this is all a mute point.
> Nope it's not just your system, well, at least I have heard other people
> having crash in kfilewidget at startup.
>
> As for the fancy things we do:

Well, the code you've posted doesn't trigger a crash in a simple KDE4 app,
so there's something else that triggers the problem. I'm attaching the
sample app as its pretty small (app generated from kapptemplate).

That said if I understood correctly I should just start kword and it
crashes? Well it does, backtrace attached. And I think its quite clear now
why it crashes: The kio-job in the KFileWidget constructor causes an update
of the GUI, hence showEvent is called. Unfortunately the KDirOperator isn't
created yet at this point, but please read on.

I don't understand though why the nested event loop causes GUI events to be
sent. As far as I know it shouldn't. And as I've recently seen a similar
problem with KJob (the DeferredDelete event, that was posted to the
main-app-loop got executed during a nested event loop and then also during
the next run of the main event loop) we might be facing a Qt bug here. Did
this crash start to happen "recently", i.e. after the 4.4.2 update to
qt-copy?

What I also don't understand is why I'm getting an "unknown protocol
kfiledialog" with the sample application, but not with koffice, so that
might be the reason why the sample app doesn't crash.

Andreas

--
You will have long and healthy life.


>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<


------=_Part_63560_15992956.1224330458158-- --===============1616506449== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe << --===============1616506449==--