fix clang-tidy CI job

[v4hn]

#2790 and #2674 currently show that the clang-tidy job broke, likely due to 55886f7 .

Recovering the clang-tidy job might fail for a variety of reasons, so I will mark this as WIP.
Anyone, feel free to fix and merge this if you have time.

[codecov]

Codecov Report

Merging #2792 (7cb6abb) into master (8237887) will increase coverage by 0.01%.
The diff coverage is n/a.

❗ Current head 7cb6abb differs from pull request most recent head 2d4d1fe. Consider uploading reports for the commit 2d4d1fe to get more accurate results

@@            Coverage Diff             @@
##           master    #2792      +/-   ##
==========================================
+ Coverage   60.45%   60.45%   +0.01%     
==========================================
  Files         366      366              
  Lines       31673    31673              
==========================================
+ Hits        19144    19146       +2     
+ Misses      12529    12527       -2     

Impacted Files	Coverage Δ
...eit_ros/manipulation/pick_place/src/pick_place.cpp	89.48% <0.00%> (-3.15%)	⬇️
...ipulation/pick_place/src/manipulation_pipeline.cpp	71.30% <0.00%> (-0.92%)	⬇️
...on/pick_place/src/approach_and_translate_stage.cpp	73.38% <0.00%> (-0.64%)	⬇️
...dl_kinematics_plugin/src/kdl_kinematics_plugin.cpp	77.87% <0.00%> (-0.38%)	⬇️
...raint_samplers/src/default_constraint_samplers.cpp	80.72% <0.00%> (+0.30%)	⬆️
...meterization/work_space/pose_model_state_space.cpp	83.02% <0.00%> (+0.63%)	⬆️
.../ompl_interface/src/detail/constrained_sampler.cpp	59.46% <0.00%> (+16.22%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8237887...2d4d1fe. Read the comment docs.

[v4hn]

does anyone see a connection between this pr and a bullet collision test failing with bad_alloc or has seen this failure before?

[henningkayser]

I don't see anything related to bullet. Is this repeatable? Also, FCL seems to fail for the same reason now.

[v4hn]

I don't see anything related to bullet. Is this repeatable? Also, FCL seems to fail for the same reason now.

build/moveit_core/test_results/moveit_core/gtest-test_fcl_collision_detection_panda.xml: 18 tests, 0 errors, 2 failures, 0 skipped

Sorry, it is related to FCL indeed. I misread the logs before.
I can't reproduce this issue with my local system and clang 11.0.1, so this needs to be debugged in the CI environment.
@tylerjw and @rhaschke have way more experience in this, sorry 😕

[v4hn]

As a general update: - The problem relates to clang and is not present with gcc - The test breaks with clang 10,11,12 in Release builds - compiling in Debug mode fixes the problem This is getting really annoying.

[v4hn]

I just distilled my insights from late-night debugging into a -O0 to a single (though highly crucial) source file.
Everyone is welcome to pick it up from here and find the concrete optimization pass that introduces the problem or find a way to work around it in source code.

If CI succeeds now I would still propose to merge it for now with an urgent followup TODO to unblock CI.

[v4hn]

This succeeds as soon as moveit/moveit_visual_tools#96 and PickNikRobotics/rviz_visual_tools#195 are merged, so I'll remove the WIP label already.

[tylerjw]

This succeeds as soon as ros-planning/moveit_visual_tools#96 and ros-planning/moveit_visual_tools#96 are merged, so I'll remove the WIP label already.

Did you mean to link the same PR twice or is there another PR that is also needed?

[v4hn]

Did you mean to link the same PR twice or is there another PR that is also needed?

updated above, sorry for the broken c&p.

[tylerjw]

I don't like the fix of changing the optimization level but I don't have time or any better ideas at the moment. I'll merge this and create an issue for fixing it better.