Threadweaver+KJob+dialog-cancel == segfault

Hi,

first I wanted to report a crash when the commit dialog from
the svn
plugin is "ok'ed", but that one disappeared (dunno
why, maybe because
instead of kjob->start() I now do kjob->exec()).

Anyway, now kdevelop crashes when cancelling that same
commit dialog.
The dialog is non-modal and I have a slot that connects to
the
cancelClicked() signal. The bracktrace looks pretty useless
to me, I
only see threadweaver-stuff. When hitting the cancel button
there has
been 1 threadweaver job already executed, but the one for
committing is
only created.

The code in question is under
branches/work/kdevelop/rework_svnplugin,
in the files svncommitjob.cpp and svnjobbase.cpp (for the
internalJobFailed()).

The backtrace is attached. I appreciate any comments.

Andreas

PS: Mirko, CC'ing you as threadweaver is shown in the
backtrace only,
maybe you have an idea...

-- 
Be different: conform.

_______________________________________________
KDevelop-devel mailing list
KDevelop-develkdevelop.org
https://barney.cs.uni-potsdam.de/mailman/listinf
o/kdevelop-devel

On 30.09.07 19:27:09, Andreas Pakulat wrote:
> Hi,
> 
> first I wanted to report a crash when the commit dialog
from the svn
> plugin is "ok'ed", but that one disappeared
(dunno why, maybe because
> instead of kjob->start() I now do kjob->exec()).
> 
> Anyway, now kdevelop crashes when cancelling that same
commit dialog.
> The dialog is non-modal and I have a slot that connects
to the
> cancelClicked() signal. The bracktrace looks pretty
useless to me, I
> only see threadweaver-stuff. When hitting the cancel
button there has
> been 1 threadweaver job already executed, but the one
for committing is
> only created.
> 
> The code in question is under
branches/work/kdevelop/rework_svnplugin,
> in the files svncommitjob.cpp and svnjobbase.cpp (for
the
> internalJobFailed()).
> 
> The backtrace is attached. I appreciate any comments.

Fixed that one by moving the commit dialog a bit up. Looking
at the new
code, maybe the problem was that when cancelled the dialog
was deleted
too early...

Andreas

-- 
What happened last night can happen again.

_______________________________________________
KDevelop-devel mailing list
KDevelop-develkdevelop.org
https://barney.cs.uni-potsdam.de/mailman/listinf
o/kdevelop-devel

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Andreas Pakulat wrote:
...

>> The backtrace is attached. I appreciate any
comments.
> 
> Fixed that one by moving the commit dialog a bit up.
Looking at the new
> code, maybe the problem was that when cancelled the
dialog was deleted
> too early...
Good thing I just thought about it a bit and did not try to
debug 

Cheers,
Mirko.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org


iD8DBQFHANwNYSSaITCTnKURAqZrAJ9STNR23Fs0t83oe1kM7ec+ue/NAQCd
HMgK
xOYCuoa7NqmQVQRVBSdsq98=
=Y6DB
-----END PGP SIGNATURE-----

_______________________________________________
KDevelop-devel mailing list
KDevelop-develkdevelop.org
https://barney.cs.uni-potsdam.de/mailman/listinf
o/kdevelop-devel