> Yup, but: the new debian/control setup in this MP takes care of replacing
> -gnutls with -openssl. So things should be good.
I'm not sure that's sufficient. The problem happened to me when I pulled the branch, typed "cmake", got a successful build, and then had the tests blow up with no sensible indication of what was wrong.
I think the cmake configuration step should check that the run-time dependencies are satisfied. Without this, the next unlucky person to come along and do what I did will get the same surprise.
From memory, I think the python stuff also has some run-time dependencies? These should be verified by cmake too.
> Yup, but: the new debian/control setup in this MP takes care of replacing
> -gnutls with -openssl. So things should be good.
I'm not sure that's sufficient. The problem happened to me when I pulled the branch, typed "cmake", got a successful build, and then had the tests blow up with no sensible indication of what was wrong.
I think the cmake configuration step should check that the run-time dependencies are satisfied. Without this, the next unlucky person to come along and do what I did will get the same surprise.
From memory, I think the python stuff also has some run-time dependencies? These should be verified by cmake too.