Hi Julien,
On Fri, Jun 23, 2017 at 09:02:30PM +0200, Julien Yann Dutheil wrote:
> This is most strange indeed. The errors are all "time out" errors on unit
> tests which take the longest time (several min or so). Looks like some
> issues with the default value for the timeout setting in CTest... I would
> suggest to simply skip the unit tests by running "cmake
> -DBUILD_TESTING=FALSE ." . Does this solve the problem?
Hmmm, this would most probably lead to successfully built packages but I
would prefer spending some additional thoughts on this. It would
explain the issue why the package has built before but does not any more
without any visible change on "weak" architectures. I could imagine
that after stretch release the run on the build servers increased and
that possibly parallel builds are happening which lead to performance
loss for single packages.
Do you see any chance to increase the timeout settings by one order of
magnitude to test this hypothesis?
Kind regards
Andreas.
--
http://fam-tille.de
