Timo Röhling

No, it is not. It is because `src/testing/fault_injection.c` is compiled into the rcutils library here: https://github.com/ros2/rcutils/blob/9cf42fbb77caa4a729db7fbf6a3087f3e6dee439/CMakeLists.txt#L74 And said file then uses `rcutils_atomic_` operations, which in turn generate symbol references to...

No, the library is pure C++ at the moment. It would require either a reimplementation in Python, or some additional glue to expose the C++ classes in Python.

Can you post all the commands you ran to build from source, in particular the `cmake` and `make` (or `ninja`) invocation.

I have never experienced this type of build failure, but Google came up with [this issue](https://github.com/libcheck/check/issues/172#issuecomment-415178221), which sounds like something that could affect you.

No problem, take your time. Another option would be to disable unit tests by passing `-DBUILD_TESTING=OFF` to cmake, which will skip building libcheck completely.

> At least this issue answers my question if reloading the service is necessary ;-) Adding this to the documentation would be helpful meanwhile. On larger mail servers the domains...

Off the top of my head, the fact that the SRS MTA treats `srsdomain.tld` as a relay domain looks fishy. If I understand your setup correctly, one of two things...

Did you solve the issue? If yes, I'm curious to know what the problem was.

Try passing `-DFETCHCONTENT_TRY_FIND_PACKAGE_MODE=ALWAYS` to cmake

I'm not making any promises, but I added some support to build and link PostSRSd with musl. It seems to work okay except for the milter support, because the Sendmail...