Hi Andreas, Le 09/11/2021 à 18:29, Andreas Tille a écrit :
Hi Pierre,Am Tue, Nov 09, 2021 at 06:02:39PM +0100 schrieb Pierre Gruet:If you are not able to reproduce I can send a full build log. It would be fine for me if you would just fix - build - upload yourself. I just wanted to help a bit. ;-)Sure, thanks for your help on this! Unfortunately I am not able to reproduce this failure at home, so I would enjoy getting your build log please.Sent via private mail to not bloat the list archive with binary files.
Thanks! I saw the test failing on your side is a thread safety test. DO you have any idea why it fails on your machine? Are you enabling some multithread features by default...? Just trying to guess.
I still have in mind some failures in libsis-jhdf5-java, which were caused by a tricky garbage collector problem showing up on some machines only. We eventually got this fixed, but I truly hope it will be far easier with igv.Smells like excluding the test is a sensible thing. On the other hand if Salsa-CI and autobuilders would be OK I'd be fine if you do a source only upload.
I should say I am hesitating to remove the test about thread safety, and it is true things are OK on Salsa-CI and on the autobuilders. Yet, before source-only-uploading, I would like to at least dig a bit on the issue you are facing. But if we cannot reproduce it, then I think a source-only upload will be OK.
Kind regards Andreas.
Best regards, -- Pierre