[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: libreoffice C++ Unit tests failing since gcc 9.2.1-12 ((Failure instantiating exceptionprotector)



Hi,

On Mon, Oct 28, 2019 at 10:39:37PM +0100, Matthias Klose wrote:
> On 28.10.19 22:17, Paul Gevers wrote:
> > Dear all,
> > 
> > The visible progress on this bug report stopped several days ago. I'd
> > like to try an get it a bit further. I'm expecting frustration on all
> > sides,
> 
> yes, and side note that I will use the same terms of "several days ago" for
> a three day silence including two non-work days.
> 
> > but let's try to work together to fix the current situation.
> 
> my moreinfo tag was removed, and I'm not interested in a bts war, which rene

Because you got the needed info. "Run debian/tests/smoketest". I don't
have more info, and simply don't have time or knowledge on C++
exceptions to handle this either way.

> likes to start very often. and, no, I won't start citing rene's private

*You* reassigned the bug back with "with what rationale" and failed
to read the bug log that I mentioned the gcc upload and the exception
failure and still insisted on Dmitrys dates in the subject which I said
to be non-true in the first message of the bug.

You started the BTS war, not me. If you read the bug you could have just
kept it at gcc without any BTS war.

> > Matthias, did you have time to look into the issue? Have you discovered
> > anything that is worth knowing already? If not, do you intent to work on
> > this soon. I noticed you uploaded a new version that doesn't address
> > this RC bug, for the reason I mentioned above, could you please refrain
> > from uploading new versions unless critical issues that aren't present
> > in testing are fixed in those uploads until a version of gcc-9 migrates?
> 
> I asked for a test case, not a multi-hour build and a multi-hour test run.

And I said it exceeds my knowledge.

debian/tests/smoketest is not a "multi hour test run". It builds part of LO,
yes, but it definitely is not multi hour. It is actually ONE test.

See the script.

The autopktest actually times itself out when it takes too long (which doesn't happen
even on older hardware.)

> Sure I can start looking at this myself, but I don't have the LO domain
> knowledge anymore.  Yes, I could start bisecting, however that doesn't sound

Neither do I. I just maintain and build this. Let alone time-wise...
Regards.

Rene


Reply to: