https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125012
--- Comment #4 from Paul Thomas <pault at gcc dot gnu.org> --- Created attachment 65375 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=65375&action=edit New failure This new attachment adds a new wrinkle beyond that of comment #1. Subroutines comment1 and comment4 encapsulate them. I cannot see any way to fix them without multiple pass compilation. Both ifx and flang compile this testcase and produce the expected runtime output. Type inference in primary.cc and resolve.cc was already pushing out the boat, even with the derived type visible. It is just possible that building a new, artificial type might be persuaded to work. For now, I propose to submit the fix for comment #1 and to keep this PR open, a defer for the attached to another time. Paul
