Seth R. Johnson
Seth R. Johnson
Not quite the ideas that we discussed with @LSchwiebert today, but at least the same part of the code with the same goal.
@Y-- the `ACTIONS_RUNNER_FORCE_ACTIONS_NODE_VERSION: node16` seems not to be needed: at least, not for the case where our failure was due to running on a CentOS7 image.
Note that `G4StackManager::PushOneTrack` assumes ownership of the track pointer, so the new track reconstructor should return a `unique_ptr`.
See #1562, #1601, #1632, #1735, #1739, #1862, #1953, #1976
@pcanal I've seen this before; it's due to vecgeom 2 crashing during destruction :( If you run through gdb/lldb you'll see
@pcanal I'm working on cleaning up the issues. Did you confirm whether or not this is a crash during VecGeom teardown?
Sooo a heisenbug?
> 1. Why is the tests issuing error messages and assert? The status checker test can only test the failure mechanisms: i.e., in the host test it makes sure we...
Ok. Easy solution, VecGeom free shouldn't assert. In fact it should throw and then VecGeom destructors should catch it.
@pcanal This should be fixed by https://gitlab.cern.ch/VecGeom/VecGeom/-/merge_requests/1305.