From hirchert@atmos.uiuc.edu Tue May 15 20:04:37 2001 Received: from atmos.uiuc.edu (gale.atmos.uiuc.edu [128.174.80.128]) by dkuug.dk (8.9.2/8.9.2) with ESMTP id UAA47390 for ; Tue, 15 May 2001 20:04:36 +0200 (CEST) (envelope-from hirchert@atmos.uiuc.edu) Received: from [128.174.80.212] (account hirchert HELO hirchert.atmos.uiuc.edu) by atmos.uiuc.edu (CommuniGate Pro SMTP 3.4.6) with ESMTP-TLS id 354458 for SC22WG5@dkuug.dk; Tue, 15 May 2001 12:04:58 -0500 Message-Id: <5.1.0.14.2.20010515125739.00aa73e8@mail.atmos.uiuc.edu> X-Sender: hirchert@mail.atmos.uiuc.edu X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Tue, 15 May 2001 13:04:58 -0500 To: SC22WG5@dkuug.dk From: Kurt W Hirchert Subject: Re: (SC22WG5.2065) Interp F90/204 In-Reply-To: <200105151510.RAA46524@dkuug.dk> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed At 04:13 PM 5/15/2001 +0100, John Reid wrote: > > > > Am I correct that in the current and draft standards users cannot write > > procedures whose arguments have the semantics of intrinsic assignment, and > > hence cannot write a MVBITS of their own? If I am correct, has there > been any > > request for this capability? > >It is permitted to overload assignment for a derived type, in which >case default assignment is not available for that type. I do not >see how this is relevant to MVBITS. Yes, a user can write a MVBITS of >his or her own. > >John. I disagree. A user cannot write a MVBITS of his or her own in standard Fortran. Whatever we decide it means to say that TO and FROM may be the same variable, this is some kind of exception to the rules in 12.4.1.6, and there is no way to achieve this exception in standard Fortran. (On the other hand, the caller can always write FROM=(var) instead of FROM=var, so it is relatively easy to work around 12.4.1.6 other ways.) -Kurt -- Kurt W Hirchert hirchert@atmos.uiuc.edu UIUC Department of Atmospheric Sciences +1-217-265-0327