Robert Ramey wrote:
The CI for the boost serialization does not and never has worked. This is easy to veryify just by looking at any of the CI output for this library. And this demonstrable fact has been pointed out numerous times. This situation is never addressed.
Looking at your last GHA build on develop https://github.com/boostorg/serialization/actions/runs/6845902041 which for some reason only includes Windows jobs, there's a failure on gcc, caused by the fact that the unqualified call to `visit` at https://github.com/boostorg/serialization/blob/a20c4d97c37e5f437c8ba78f29683... finds std::visit via ADL and tries to instantiate it, which fails. This looks like a bug in libstdc++ 8 to me, as std::visit is required to SFINAE, but it could just be that the standard at the time didn't require it. This is not in any way a failure of CI - test_variant doesn't compile under GCC 8, so it tells you so. What else do you expect to happen? And who and how is supposed to "address this situation"?