> Thus I believe the testcase exposes a pre-existing bug in gfortran. What > should we do? XFAIL the testcase and open a PR? Yes, assuming you can't immediately fix the problem :-) Paul