On Monday 18 July 2011 21:37:24 Tobias Burnus wrote:
> As the coarray status is nontrivial to check, I have created a function
> for that - and I use it now for the interface checking. There are more
> cases, where the wrong check is used, leading to accepts-invalid and
> rejects-valid issues; however, I would like to track them later.
Two minor suggestions:
> diff --git a/gcc/fortran/expr.c b/gcc/fortran/expr.c
> index b8eb555..67ac2e5 100644
> --- a/gcc/fortran/expr.c
> +++ b/gcc/fortran/expr.c
> @@ -4154,6 +4154,74 @@ gfc_is_coindexed (gfc_expr *e)
> }
>
>
> +/* Coarrays are variables with a corank but not being coindexed. However,
also
> + the following is a coarray: A subobject of a coarray is a coarray if it
does
> + not have any cosubscripts, vector subscripts, allocatable component
> + selection, or pointer component selection. (F2008, 2.4.7) */
> +
> +bool
> +gfc_is_coarray (gfc_expr *e)
> +{
> + gfc_ref *ref;
> + gfc_symbol *sym;
> + gfc_component *comp;
> + bool coindexed;
> + bool coarray;
> + int i;
> +
> + if (e->expr_type != EXPR_VARIABLE
> + && e->expr_type != EXPR_FUNCTION)
> + return false;
This should not change anything, but I think EXPR_FUNCTION can be discarded
because of C523:
A coarray shall be a component or a variable that is not a function result.
> diff --git a/gcc/fortran/interface.c b/gcc/fortran/interface.c
> index dcf6c4e..fa23de7 100644
> --- a/gcc/fortran/interface.c
> +++ b/gcc/fortran/interface.c
> @@ -1557,41 +1557,21 @@ compare_parameter (gfc_symbol *formal, gfc_expr
*actual,
[...]
> + if (formal->attr.codimension && formal->attr.allocatable)
> + {
> + gfc_ref *last = NULL;
>
> for (ref = actual->ref; ref; ref = ref->next)
> - {
> - if (ref->type == REF_ARRAY && ref->u.ar.as->corank
> - && ref->u.ar.type != AR_FULL && ref->u.ar.dimen != 0)
> - {
> - if (where)
> - gfc_error ("Actual argument to '%s' at %L must be a coarray
"
> - "and thus shall not have an array designator",
> - formal->name, &ref->u.ar.where);
> - return 0;
> - }
> - if (ref->type == REF_COMPONENT)
> - last = ref;
> - }
> + if (ref->type == REF_COMPONENT)
> + last = ref;
>
> /* F2008, 12.5.2.6. */
> if (formal->attr.allocatable &&
This formal->attr.allocatable condition is now superfluous.
Now about your request:
> I would be happy if someone could carefully read the function in expr.c
> - and confirm that the second test case ("dg-error" line) is indeed
> invalid. I think it is invalid because one has an array element of a
> coarray, which is a coarray but which is not simply contiguous.
I don't understand what is your concern here; the standard seems pretty clear:
12.5.2.8 (f2008):
If the dummy argument is an array coarray that has the CONTIGUOUS
attribute or is not of assumed shape, the corresponding actual argument
shall be simply contiguous.
Thus, invalid indeed.
> Thus, if possible cross check that it is indeed not simply contiguous.
6.5.4 (f2008):
A section-subscript-list specifies a simply contiguous section if and only if
it does not have a vector subscript and
• all but the last subscript-triplet is a colon,
• the last subscript-triplet does not have a stride, and
• _no subscript-triplet is preceded by a section-subscript that is a_
_subscript._
It's not simply contiguous indeed. Wasn't it obvious anyway or did I miss
something?
So, to sum up, if one sticks to fortran 2008, the test is clearly invalid
and...
> Build and regtested on x86-64-linux.
> OK for the trunk?
... the patch is ok with the two minor changes suggested above.
Mikael
