gcc/testsuite/gfortran.dg/dependent_decls_1.f90

   1 ! { dg-do run }
   2 ! Tests the fix for pr28660 in which the order of dependent declarations
   3 ! would get scrambled in the compiled code.
   4 !
   5 ! Contributed by Erik Edelmann  <erik.edelmann@iki.fi>
   6 !
   7 program bar
   8     implicit none
   9     real :: x(10)
  10     call foo1 (x)
  11     call foo2 (x)
  12     call foo3 (x)
  13 contains
  14     subroutine foo1 (xmin)
  15         real, intent(inout) :: xmin(:)
  16         real :: x(size(xmin)+1)           ! The declaration for r would be added
  17         real :: r(size(x)-1)              ! to the function before that of x
  18         xmin = r
  19         if (size(r) .ne. 10) STOP 1
  20         if (size(x) .ne. 11) STOP 2
  21     end subroutine foo1
  22     subroutine foo2 (xmin)                ! This version was OK because of the
  23         real, intent(inout) :: xmin(:)    ! renaming of r which pushed it up
  24         real :: x(size(xmin)+3)           ! the symtree.
  25         real :: zr(size(x)-3)
  26         xmin = zr
  27         if (size(zr) .ne. 10) STOP 3
  28         if (size(x) .ne. 13) STOP 4
  29     end subroutine foo2
  30     subroutine foo3 (xmin)
  31         real, intent(inout) :: xmin(:)
  32         character(size(x)+2) :: y         ! host associated x
  33         character(len(y)+3) :: z          ! This did not work for any combination
  34         real :: r(len(z)-5)              ! of names.
  35         xmin = r
  36         if (size(r) .ne. 10) STOP 5
  37         if (len(z) .ne. 15) STOP 6
  38     end subroutine foo3
  39 end program bar