From owner-sc22wg5@open-std.org Tue Dec 9 13:24:38 2008 Return-Path: X-Original-To: sc22wg5-dom7 Delivered-To: sc22wg5-dom7@www2.open-std.org Received: by www2.open-std.org (Postfix, from userid 521) id 8F318CA343D; Tue, 9 Dec 2008 13:24:38 +0100 (CET) X-Original-To: sc22wg5@open-std.org Delivered-To: sc22wg5@open-std.org Received: from ppsw-6.csi.cam.ac.uk (ppsw-6.csi.cam.ac.uk [131.111.8.136]) by www2.open-std.org (Postfix) with ESMTP id 0A3CACA3439 for ; Tue, 9 Dec 2008 13:24:37 +0100 (CET) X-Cam-AntiVirus: no malware found X-Cam-SpamDetails: not scanned X-Cam-ScannerInfo: http://www.cam.ac.uk/cs/email/scanner/ Received: from hermes-1.csi.cam.ac.uk ([131.111.8.51]:52653) by ppsw-6.csi.cam.ac.uk (smtp.hermes.cam.ac.uk [131.111.8.156]:25) with esmtpa (EXTERNAL:nmm1) id 1LA1dl-0001pt-K0 (Exim 4.70) for sc22wg5@open-std.org (return-path ); Tue, 09 Dec 2008 12:24:37 +0000 Received: from prayer by hermes-1.csi.cam.ac.uk (hermes.cam.ac.uk) with local (PRAYER:nmm1) id 1LA1dl-0007kW-5y (Exim 4.67) for sc22wg5@open-std.org (return-path ); Tue, 09 Dec 2008 12:24:37 +0000 Received: from [131.111.10.32] by webmail.hermes.cam.ac.uk with HTTP (Prayer-1.3.1); 09 Dec 2008 12:24:37 +0000 Date: 09 Dec 2008 12:24:37 +0000 From: "N.M. Maclaren" To: WG5 Subject: N1761, TYPE(*), BIND(C) and arrays Message-ID: In-Reply-To: <20081202130345.82E16C178E1@www2.open-std.org> References: <20081127193527.EF00DC178D9@www2.open-std.org> <20081202130345.82E16C178E1@www2.open-std.org> X-Mailer: Prayer v1.3.1 Mime-Version: 1.0 Content-Type: text/plain; format=flowed; charset=ISO-8859-1 Sender: owner-sc22wg5@open-std.org Precedence: bulk This is something that I subconsciously realised when reading N1761, but only formulated what it was fairly recently. Fortran does not have the concept of a called procedure knowing anything about the properties of its actual arguments (i.e. the objects and expressions in the calling procedure). The SOLE information available to a called procedure about its arguments are in its OWN declaration (i.e. the properties of its dummy arguments). It overlaps with Reinhold's points to a considerable extent, of course. That makes TYPE(*) effectively non-interoperable with the proposal to have two incompatible argument passing mechanisms for BIND(C), which are selected solely on the types of argument. Here is an example program. Yes, some of the examples may already be forbidden - I don't have time to check - and there may be other bugs. But, for N1761 to be acceptable as a TR, all of these cases have to be specified or stated to be processor dependent. And please remember that the TARGET and POINTER attributes have 'interesting' properties as far as when aliasing and update are allowed. MODULE Mudpuddle INTERFACE SUBROUTINE Slough (arg) BIND(C) TYPE(*), DIMENSION(..) :: arg END SUBROUTINE Slough END INTERFACE END MODULE Mudpuddle MODULE Morass REAL :: array_A(10) = 0.0 END MODULE Morass PROGRAM Swamp USE Mudpuddle USE Morass REAL :: array_B(10) = 0.0 REAL, ALLOCATABLE :: array_C(:) REAL, POINTER :: array_D(:) REAL, TARGET :: array_E(10) = 0.0 INTEGER, PARAMETER :: index_P(10) = (/ (n, n = 1,10) /), & index_Q(3) = (/ 3, 4, 5, 6, 7 /), index_R = (/ 3, 5, 7 /) ALLOCATE(array_C(10)) array_D => array_E ! For each of the following calls, is the array passed using descriptors ! or not, why, and where is that stated? Note that many of those calls ! use expressions that just happen to deliver a contiguous section, and ! so an optimising compiler could use either mechanism. CALL Slough(array_A) CALL Slough(array_A(:)) CALL Slough(array_A(3:7)) CALL Slough(array_A(3:7:2)) CALL Slough(array_A(index_P)) CALL Slough(array_A(index_Q)) CALL Slough(array_A(index_R)) CALL Slough(array_B) CALL Slough(array_B(:)) CALL Slough(array_B(3:7)) CALL Slough(array_B(3:7:2)) CALL Slough(array_B(index_P)) CALL Slough(array_B(index_Q)) CALL Slough(array_B(index_R)) CALL Slough(array_C) CALL Slough(array_C(:)) CALL Slough(array_C(3:7)) CALL Slough(array_C(3:7:2)) CALL Slough(array_C(index_P)) CALL Slough(array_C(index_Q)) CALL Slough(array_C(index_R)) CALL Slough(array_D) CALL Slough(array_D(:)) CALL Slough(array_D(3:7)) CALL Slough(array_D(3:7:2)) CALL Slough(array_D(index_P)) CALL Slough(array_D(index_Q)) CALL Slough(array_D(index_R)) CALL Slough(array_E) CALL Slough(array_E(:)) CALL Slough(array_E(3:7)) CALL Slough(array_E(3:7:2)) CALL Slough(array_E(index_P)) CALL Slough(array_E(index_Q)) CALL Slough(array_E(index_R)) END PROGRAM Swamp Regards, Nick Maclaren, University of Cambridge Computing Service, New Museums Site, Pembroke Street, Cambridge CB2 3QH, England. Email: nmm1@cam.ac.uk Tel.: +44 1223 334761 Fax: +44 1223 334679