https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38319
--- Comment #10 from Paul Thomas <pault at gcc dot gnu.org> --- Of the testcases originally highlighted: alloc_comp_assign_2.f90 still leaks alloc_comp_assign_4.f90 still leaks alloc_comp_basics_2.f90 still leaks alloc_comp_basics_5.f90 is OK - probably always was! alloc_comp_constructor_2.f90 still leaks alloc_comp_initializer_1.f90 still leaks alloc_comp_constructor_2.f90 also leaks It's time to deal with this! Paul