Hello,

I applied the patch. The change will be part of the next release. However, I 
haven't had this issue 
on my openSUSE system (13.1) over here. Can this be a timing issue as we have 
seen in some 
other situations were C++ objects were destroyed beneath the still existing 
Python instance?

Detlev

On Thursday 27 March 2014, 00:12:03 Hans-Peter Jansen wrote:
> Dear Detlev,
> 
> I'm too silly for words. Phil gave the right hint, when I reported this
> issue in December just a few hours after report.. I must have missed his
> answer, only to rediscover sip.setdestroyonexit myself lately, but couldn't
> get its argument right :(. Oh well, that's the result of doing too many
> things in parallel..
> 
> Detlev, would you please apply attached patch for eric4, and all is fine.
> 
> Phil, Detlev, sorry for the churn. I feel really bad about this.
> 
> Rohan, I bet, adding a similar sequence to apport early in the execution
> path will fix it as well.
> 
> Cheers,
> Pete
> 
> P.S.: Could get rid of two reproducible crashes in about 24 hours, *great*.
> 
> On Mittwoch, 26. März 2014 20:14:06 Hans-Peter Jansen wrote:
> > [Reposted on Detelv's request in order to address also the qscintilla ML
> > --- Note to Phil: I really think, that sip.setdestroyonexit is the right
> > approach to fix this issue (class), the question is, why it doesn't..]
> > 
> > Hi Detlev,
> > 
> > you're right - splitting is a red herring in this case, just choosing a
> > "New view" is sufficient to trigger the exact same traceback, so
> > QScintilla is most likely to be in charge of is crash.
> > 
> > Thanks for the hint, Detlev.
> > 
> > Kind regards,
> > Pete-- 
*Detlev Offenbach*
[email protected]
_______________________________________________
QScintilla mailing list
[email protected]
http://www.riverbankcomputing.com/mailman/listinfo/qscintilla

Reply via email to