[RHEL-CI] Source underlay with generate_parameter_library

[christophfroehlich]

generate_parameter_library is necessary for #971, and somehow it is not released for RHEL: This project is currently disabled

For build details see
https://github.com/ros-controls/ros2_rhel/blob/7bfd5933856388cbfee493e1805b6d95cfee32f6/Dockerfile.rhel9#L49-L61
https://github.com/ros-controls/ros2_rhel/blame/7bfd5933856388cbfee493e1805b6d95cfee32f6/Dockerfile.rhel8#L33-L46

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (5b5f332) 47.53% compared to head (a8b05a2) 47.53%.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1338   +/-   ##
=======================================
  Coverage   47.53%   47.53%           
=======================================
  Files          41       41           
  Lines        3549     3549           
  Branches     1931     1931           
=======================================
  Hits         1687     1687           
  Misses        459      459           
  Partials     1403     1403           

Flag	Coverage Δ
unittests	47.53% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

[destogl]

Should RHEL CI still fail?

[bmagyar]

well now it fails with compiling messages 🤷

[christophfroehlich]

well now it fails with compiling messages 🤷

Something is broken with the empy module, I had to fix it to a certain version and thought I tried it locally. I'll work on that again next week..

[bmagyar]

Happy to leave this PR open for the empy fixes too? I don't see a point of a merge train which will need another merge train to really fix

[christophfroehlich]

The RHEL job builds now rather fine, but there are consistently flaky segfaulting tests ResourceManagerTest.initialization_with_wrong_urdf or ResourceManagerTest.load_urdf_called_if_urdf_is_invalid, see here with ubuntu semi-binary build. I was able to reproduce this locally without giving an explicit build type to colcon, but in the CI it also fails with rel-with-deb-info.

@bmagyar @destogl IMHO this is not related to this PR any more, but a general issue.

[fmauch]

Seems to be fair and in line with the Docker image.

[bmagyar]

Thanks for the summary