https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85599
--- Comment #7 from janus at gcc dot gnu.org --- (In reply to janus from comment #4) > What worries me most is that other compilers choose different > implementations. Can this kind of code really be processor-dependent? Most other compilers that I tried actually behave different than gfortran, calling 'check' twice in the test case of comment #0 (ifort, flang, PGI). The only compiler that behaves like gfortran and optimizes the second call out is the Oracle compiler (sunf95). All gfortran versions since at least 4.4 behave in the same way.
