https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108680
--- Comment #6 from Jerry DeLisle <jvdelisle at gcc dot gnu.org> ---
The check is being done in interface.cc. The kind is being checked against
default_integer_kind.
case(2): /* UNIT */
type = BT_INTEGER;
kind = gfc_default_integer_kind;
intent = INTENT_IN;
check_dtio_arg_TKR_intent (fsym, typebound, type, kind,
0, intent);
break;
case(3): /* IOTYPE */
type = BT_CHARACTER;
kind = gfc_default_character_kind;
intent = INTENT_IN;
check_dtio_arg_TKR_intent (fsym, typebound, type, kind,
0, intent);
break;
I believe they are correct. Going back to my example:
---
subroutine write_formatted (this, unit, iotype, v_list, iostat, iomsg)
class(dtype), intent(in) :: this
integer(8), intent(in) :: unit, v_list(:)
---
Explicitly giving the kind=8 in the interface specification violates the
interface definition given in the standard. If I compile it like this, the
errors go away as expected.
$ gfc -fdefault-integer-8 pr108680.f90
jerry@r7laptop:~/dev/test/pr108680$ ./a.out
At line 21 of file pr108680.f90
Fortran runtime error: Unit number in I/O statement too large
In the runtime library we assume a kind=4 integer in transfer.c.
case FMT_DT:
if (n == 0)
goto need_data;
GFC_INTEGER_4 unit = dtp->u.p.current_unit->unit_number;
char dt[] = "DT";
char tmp_iomsg[IOMSG_LEN] = "";
char *child_iomsg;
gfc_charlen_type child_iomsg_len;
GFC_INTEGER_4 noiostat;
Do we need to support more units than we can specify with GFC_INTEGER_4?
