[PATCH, Fortran] PROCEDURE declarations

Tobias Burnus burnus@net-b.de
Sun Sep 2 09:05:00 GMT 2007


Tobias Schlüter wrote:
>>
>> +      gfc_error ("Procedure pointers used at %L are "
>> +        "not yet implemented", where);
>
> The wording should make very explicit that this is a compiler
> deficiency, maybe use "sorry"?
I think "not implemented" should be clear enough, isn't it?

> I also feel that this is an omission that limits the usefulness of
> this language feature by quite a lot, but I only started reading about
> F2K's features, so I may be misestimating.
Well, the idea was to have PROCEDURE without pointer first and implement
procedure pointer next. (Janus is already working on this.)

But PROCEDURE by itself is already useful; see e.g.
http://groups.google.com/group/comp.lang.fortran/browse_thread/thread/4d51d6ca89f7d4f8/

>> +      gfc_error ("Intrinsic procedure '%s' not yet supported "
>> +            "in PROCEDURE statement at %C", proc_if->name);
>
> Again, make clear that this is a compiler deficiency.  People may
> wonder if "not yet supported" may mean "at a later point in the
> program, this would work".

Would be "not yet implemented" better?

> Once Tobias B believes the language support is complete, I don't doubt
> that it will be ok from my POV.
I think PROCEDURE is complete except of the known TODOs: (a)
"Inheriting" the interface from intrinsic procedures, (b) procedure
pointers and (c) PROCEDURE in/as type-bound procedure. At least I could
not find anymore find a test program which is mishandled.

Of these, (b) is planned for the near future; Janus also planned to look
at (c).

Tobias



More information about the Gcc-patches mailing list