[J3] Is this standard-conforming?

[Vipul Parekh]

Upon further reading of 10-007r1 document, I think the last sentence of
paragraph 1 in section 16 5.1.14 Host association is what applies here
which has:

"The accessed entities are named data objects, derived types, .."

and since a named constant is nothing but a named data object with the
PARAMETER attribute, I now think Steve is correct and the code is
conforming.

Please ignore my earlier note on this.

Regards,
Vipul

On Thu, Jul 26, 2018, 10:51 AM Vipul Parekh <parekhvs at gmail.com> wrote:

> I think the code is non-conforming.  This is on account of an
> inaccessible entity in terms of host association,
>
> With 10-007r1 document toward Fortran 2008, section 12.4.3.3 IMPORT
> statement has in paragraph 1,
> "The name of an entity made accessible by an IMPORT statement shall
> not appear in any of the contexts described in 16.5.1.4 that cause the
> host entity of that name to be inaccessible."
>
> Then in paragraph 1 in section 16.5.1.4 Host association,
> "An interface body that is not a separate interface body has access
> via host association to the named entities from its host that are made
> accessible by IMPORT statements in the interface body."
>
> And in paragraph 2,
> "A name that appears in the scoping unit as .. a named-constant in a
> named-constant-def in a parameter-stmt .. is a local identifier in the
> scoping unit and any entity of the host that has this as its
> nongeneric name is inaccessible by that name by host association."
>
> I might be wrong but my interpretation of above is the named constant
> c_name is inaccessible in the interface body.
>
> I'll be keen to read further comments on this.
>
> Regards,
> Vipul
> On Thu, Jul 26, 2018 at 9:25 AM Steve Lionel via J3
> <j3 at mailman.j3-fortran.org> wrote:
> >
> > On 7/26/2018 6:12 AM, Damian Rouson via J3 wrote:
> > > One processor accepts the code below.  Three others generate error
> > > messages related to the binding label c_name and either report a
> > > syntax error or report that the binding label is a variable that does
> > > not reduce to a constant expression.
> >
> > It is clear to me that this is standard-conforming. The binding label is
> > required to be a default-char-constant-expression, and c_name, visible
> > in the interface block through the import, is exactly that. If there is
> > some additional restriction on the NAME= value that prohibits this, I
> > can't find it.
> >
> > As a former developer, though, I can understand how this sort of thing
> > happens.
> >
> > Steve
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20180726/d8d2ded4/attachment.html>