This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [gfortran] Fix PR15963
Paul Brook wrote:
>>2004-06-22 Tobias Schlueter <tobias.schlueter@physik.uni-muenchen.de>
>>
>> PR fortran/15963
>> * expr.c (check_intrinsic_op): Allow comparison of characters.
>> Make logic easier.
>
>
> Capitalise CHARACTER in the error message.
> Otherwise, Ok.
I committed this in the form shown below. I slightly revised the check,
to put it in a more readable form, as I had messed up when applying my
patch to a clean tree, and after it had taken me a while to untangle the
logical expression, to see where the bug was, I decided to make up a
clearer one.
- Tobi
Index: expr.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/fortran/expr.c,v
retrieving revision 1.7
diff -u -p -r1.7 expr.c
--- expr.c 27 May 2004 12:35:11 -0000 1.7
+++ expr.c 29 Jun 2004 16:59:48 -0000
@@ -1167,6 +1167,17 @@ check_intrinsic_op (gfc_expr * e, try (*
case INTRINSIC_GE:
case INTRINSIC_LT:
case INTRINSIC_LE:
+ if ((*check_function) (e->op2) == FAILURE)
+ return FAILURE;
+
+ if (!(et0 (e->op1) == BT_CHARACTER && et0 (e->op2) == BT_CHARACTER)
+ && !(numeric_type (et0 (e->op1)) && numeric_type (et0 (e->op2))))
+ {
+ gfc_error ("Numeric or CHARACTER operands are required in "
+ "expression at %L", &e->where);
+ return FAILURE;
+ }
+ break;
case INTRINSIC_PLUS:
case INTRINSIC_MINUS:
@@ -1179,10 +1190,8 @@ check_intrinsic_op (gfc_expr * e, try (*
if (!numeric_type (et0 (e->op1)) || !numeric_type (et0 (e->op2)))
goto not_numeric;
- if (e->operator != INTRINSIC_POWER)
- break;
-
- if (check_function == check_init_expr && et0 (e->op2) != BT_INTEGER)
+ if (e->operator == INTRINSIC_POWER
+ && check_function == check_init_expr && et0 (e->op2) !=
BT_INTEGER)
{
gfc_error ("Exponent at %L must be INTEGER for an initialization "
"expression", &e->op2->where);