TST: gdb has been removed from Travis #10216
Description
Bug report
Travis builds were failing. See below. Easy kludge was to remove the gdb calls before the pytest call as in #10217. Passes test, but C++ will no longer get a stacktrace if there are errors.
This is a placeholder to remind us that gdb should really still be called and the issue investigated.
OTOH, feel free to close if we don't think the gdb traces are actually crucial.
Passes
https://travis-ci.org/matplotlib/matplotlib/jobs/325962082
if [[ $TRAVIS_OS_NAME == 'osx' ]]; then
pytest $PYTEST_ARGS $RUN_PEP8
else
gdb -return-child-result -batch -ex r -ex bt --args python $PYTHON_ARGS -m pytest $PYTEST_ARGS $RUN_PEP8
fi
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
============================= test session starts ==============================
Fails
https://travis-ci.org/matplotlib/matplotlib/jobs/327083036
if [[ $TRAVIS_OS_NAME == 'osx' ]]; then
pytest $PYTEST_ARGS $RUN_PEP8
else
gdb -return-child-result -batch -ex r -ex bt --args python $PYTHON_ARGS -m pytest $PYTEST_ARGS $RUN_PEP8
fi
warning: Error disabling address space randomization: Operation not permitted
============================= test session starts ==============================
@QuLogic also notes that the kernel is different between both tests (though they updated their kernel due to Specter/Meltdown, so maybe all the kernels are new).
Not sure what the solution is. See