Hi Jochen. Thanks for getting back to me this fast. Interestingly, I had the exact same error as the reproducible build log I linked in https://en.wikipedia.org/wiki/GLinux (well, an intermediate state of it) that pretty accurately mirrors Debian testing, except that we rebuild all packages. I can't currently reproduce in pure debian-testing (though I thought I had done that before, not sure anymore). I'll have a look at differences in the build dependencies which are usually what triggers such different results.
Meanwhile, if you have good knowledge of the source code: Any idea why the return code from "rosrun tf2_ros static_transform_publisher" might be 3 instead of -1 (aka 255) in the three test cases in https://salsa.debian.org/science-team/ros-geometry2/-/blob/master/test_tf2/test/test_static_publisher.py?ref_type=heads - because I honestly was unable to find that. Feel free to close the report. If I have something more substantial (i.e. a more clear reproduction case), I'll update and reopen. Danke, Sven
