Pyroot gSystem.Load segfault (pt2)

[bleu65]

Continuation of Pyroot gSystem.Load segfault which was automatically closed and can’t be reopened apparently.

Thanks @etejedor! So, I got several friends and group members to try. For 5 different people it segfaulted and always does so. For you and 1 other person it always works.

The python load is pretty simple and the shared library is definitely there. There’s nothing in our environments - just the default lxplus (with bash) setup.

Do you know why it’s segfaulting? The only thing different from our side is that the files are on cvmfs. Could it be some file access problem do you think?

Presumably a segfault is a bug in ROOT though as a segfault just shouldn’t happen. ie no exception or error message or return -1 status.

For the jitting of the header comment - I’m not so sure. I’ve never used gInterpreter.Declare in our code before and it works elsewhere fine. Also, it would seem to work for you and another user just fine. I do do gSystem.AddIncludePath(os.environ[‘ROOT_INCLUDE_PATH’]) though before the gSystem.Load().

A bit mystical. Any ideas would be appreciated! Thanks.

[etejedor]

Hi,

Just to kick CVMFS out of the equation: could you or any of your colleagues that see the error try to copy everything elsewhere (e.g. /tmp on lxplus, or even AFS or EOS) and see if that works?

Also, what happens if you try with LCG 100, which just came out? Changing the software stack on CVMFS helps?

Another thing: is it possible that there is something in the environment of the people who see the error that is interfering? E.g. stuff installed on $HOME/.local, or any environment setting.