Hi! > This looks suspiciously like PR fortran/18108. This is a very long-standing > problem, and obviously not easy to fix :( Sorry, just saw that you are obviously aware of that PR :) So, do you think this is a different problem? Cheers, Martin