[Patch, Fortran] PR 41766: [OOP] SELECT TYPE selector as actual argument with INTENT(INOUT)

[Janus Weil <janus at gcc dot gnu dot org>]

Hi all,

the attached patch fixes another problem with SELECT TYPE and is
pretty obvious. The problem is that the temporaries which are inserted
for the TYPE IS cases miss the right flavor. The patch simply sets the
flavor to FL_VARIABLE. Moreover it uses the canonical functions for
setting attributes.

Regtested on x86_64-unknown-linux-gnu. Ok for trunk?

Cheers,
Janus


2009-10-20  Janus Weil  <janus@gcc.gnu.org>

	PR fortran/41766
	* match.c (select_type_set_tmp): Set flavor for temporary.

2009-10-20  Janus Weil  <janus@gcc.gnu.org>

	PR fortran/41766
	* gfortran.dg/select_type_7.f03: New test case.

Index: gcc/fortran/match.c
===================================================================
--- gcc/fortran/match.c	(Revision 153009)
+++ gcc/fortran/match.c	(Arbeitskopie)
@@ -4047,9 +4047,10 @@ select_type_set_tmp (gfc_typespec *ts)
 
   sprintf (name, "tmp$%s", ts->u.derived->name);
   gfc_get_sym_tree (name, gfc_current_ns, &tmp, false);
-  tmp->n.sym->ts = *ts;
-  tmp->n.sym->attr.referenced = 1;
-  tmp->n.sym->attr.pointer = 1;
+  gfc_add_type (tmp->n.sym, ts, NULL);
+  gfc_set_sym_referenced (tmp->n.sym);
+  gfc_add_pointer (&tmp->n.sym->attr, NULL);
+  gfc_add_flavor (&tmp->n.sym->attr, FL_VARIABLE, name, NULL);
 
   select_type_stack->tmp = tmp;
 }

Attachment: select_type_7.f03
Description: Binary data