Re: [patches,fortran] Add POINTER INTENT(*) support (PR29624)

[Paul Thomas <paulthomas2 at wanadoo dot fr>]

Tobias,
> > Regression tested on x86_64-unknown-linux-gnu.
> >     
OK with a couple of questions/niggles....
> >  
> > +/* Given a symbol of a formal argument list and an expression, see if
> > +   their intents are compatible.  Returns nonzero if compatible,
> > +   zero if not compatible.  */
> > +
> >     
"Given a symbol of a formal argument list and an expression, return 
non-zero if their intents are compatible, zero otherwise." is a bit more 
concise.
> > -      return gfc_notify_std (standard, "In the selected standard, %s attribute "
> > -                             "conflicts with %s attribute at %L", a1, a2,
> > -                             where);
> > +      return gfc_notify_std (standard, "%s: %s attribute with %s attribute "
> > +                             "at %L",
> > +                             (standard == GFC_STD_F2003)
> > +                               ? "Fortran 2003" : "Not in the selected standard",
> > +                             a1, a2, where);
> >  
Is the language support OK with this? I seem to recall that we had some 
problems with computed messages.
> > +  /* Check INTENT(IN), unless the object itself of higher up in the
> > +     hierarchy is a pointer.  */
> >     
Not only is the symtax a bit odd but I think that , "unless the object 
itself is the component or sub-component of a pointer." is clearer.
> > +
> > +  /* Check INTENT(IN), unless the object itself of higher up in the
> > +     hierarchy is a pointer.  */
> >     
ditto
> >  
> > +  /* Check INTENT(IN), unless an object higher up in the
> > +     hierarchy is a pointer.  */
> >     
ditto**2

Thanks!

Paul