------- Comment #3 from burnus at gcc dot gnu dot org 2008-04-16 05:05 ------- I think this is wrong: real :: z(t%n) type (tt) :: t
the order has to be reversed. The following is also invalid (with implicit none): real :: z(m) integer :: m Most compilers support this, however. I think gfortran behaves OK; though, one could argue that for consistency also the first has to be supported or that second one should be diagnosed (only with -std=, there are too many programs with reversed order). When I have time, I will search for the quote from the standard. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35951
