From J.L.Schonfelder@liverpool.ac.uk Fri Jun 24 06:33:45 1994 Received: from mailhub.liverpool.ac.uk (mail.liv.ac.uk) by dkuug.dk with SMTP id AA12206 (5.65c8/IDA-1.4.4j for ); Fri, 24 Jun 1994 14:34:18 +0200 Received: from liverpool.ac.uk by mailhub.liverpool.ac.uk with SMTP (PP) id <25937-0@mailhub.liverpool.ac.uk>; Fri, 24 Jun 1994 13:33:50 +0100 Date: Fri, 24 Jun 1994 13:33:45 PDT From: Lawrie Schonfelder Subject: Re: (SC22WG5.576) Duplicate Interface Declaration To: "Erik W. Kruyt (+31 71 276804)" Cc: SC22/WG5 members Message-Id: Priority: Normal Mime-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII X-Charset: ASCII X-Char-Esc: 29 I had thought about your question which is why I used the word identical. I mean identical literally, But if implementers could do better than that I would be happy to relax the restriction. My point still holds though The standard rightly requires interface declarations of external procedures to match those of the actual procedure with no mechanism at all to check this match. For module or internal procedures this checking can be done. On Fri, 24 Jun 1994 09:49:18 +0100 (MET) Erik W. Kruyt (+31 71 276804) wrote: > From: Erik W. Kruyt (+31 71 276804) > Date: Fri, 24 Jun 1994 09:49:18 +0100 (MET) > Subject: Re: (SC22WG5.576) Duplicate Interface Declaration > To: J.L.Schonfelder@liverpool.ac.uk > > Dear Lawry, > > Duplicating declarations can be tricky as we have learned from languages like > Modula-2. The problem is: what is identical? > Should it mean literally, syntactically or semantically? Users give preference > to the last, but that can be complicated or even ambiguous for a compiler. > You require already that dummy arguments must have the same name. But do we > allow the following to be the same? > > INTERFACE > SUBROUTINE SUB(IARG) > INTEGER, DIMENSION(10) :: IARG > END SUBROUTINE > END INTERFACE > .. > CONTAINS > SUBROUTINE SUB (IARG) > INTEGER, PARAMETER :: IS=10 > INTEGER IARG(IS) > END SUBROUTINE I would like to think a smart compiler could recognise these to be the same and accept them, but I hope I have written the proposal so as to not require the compiler to do so. INTERFACE FUNCTION FUN(IARG) INTEGER, DIMENSION(2*IARG) :: FUN END FUNCTION END INTERFACE .. CONTAINS FUNCTION FUN(IARG) INTEGER, DIMENSION(IARG+IARG) :: FUN END FUNCTION I would definitely expect the processor to flag the above. In general duplication in this area is needed to allow definitions to be done in both places where frequently different issues are the design priority (overload resolution in one and specific procedure implementation in the other. The safe and desirable maintenance requirement is that these two definitions then be kept strictly in identical step. > > Erik -- Dr.J.L.Schonfelder Director, Computing Services Dept. The University of Liverpool, UK e-mail J.L.Schonfelder@liv.ac.uk phone: +44(51)794-3716 fax: +44(51)794-3759