From j.l.schonfelder@liverpool.ac.uk Mon Aug 10 16:32:52 1998 Received: from ratatosk.DK.net (root@[193.88.44.22]) by dkuug.dk (8.6.12/8.6.12) with ESMTP id QAA00315 for ; Mon, 10 Aug 1998 16:32:51 +0200 Received: from pcmail.liv.ac.uk (exim@pcmail.liv.ac.uk [138.253.252.13]) by ratatosk.DK.net (8.8.8/8.8.8) with SMTP id NAA26611 for ; Mon, 10 Aug 1998 13:19:36 +0200 (MET DST) Received: from jlspcnt.liv.ac.uk [138.253.102.118] by pcmail.liv.ac.uk with smtp (Exim 1.73 #2) id 0z5pzA-0001vR-00; Mon, 10 Aug 1998 12:19:32 +0100 From: Lawrie Schonfelder Reply-To: j.l.schonfelder@liverpool.ac.uk To: SC22/WG5 members Subject: String standard, module and F95 Message-ID: Date: Mon, 10 Aug 1998 12:19:33 +0100 (British Summer Time) Priority: NORMAL X-Mailer: Simeon for Win32 Version 4.1.5 Build (43) X-Authentication: IMSP MIME-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII I was doing some programming over the weekend to try and implement the decision by WG5 that the SPLIT subroutine should be elemental. Much to my surprise and considerable chagrin I discovered that I could not make the procedures elemental since they needed to use in internal working string whose length was declared depending on the SIZE of a dummy argument. For some entirely unfathomable reason this is barred by the standard. At that point I all but gave up on Fortran and decided to through in my lot with Java. Why oh why must Fortran continue to be lttered with these arbitrary, IMO, unjustified restrictions. This one makes just about any data-abstraction based on dynamic types or even those that need automatic local work arrays impossible to extend elementally. The standard justifies this (adjective deleted) restriction on the grounds of optimisation. What optimisation? This smacks to me of an excuse for sloppy compilation. If I need this sort of functionality the code I would have to produce to work around this problem would be vastly less efficient in my time and I suspect also in machine time. If I dont need this functionality and dont use it I would expect a half way decent optimising compiler to be able to detect that and apply what ever optimisation was being protected. I despair! When will WG5 learn that petty restrictions like this are killing Fortran if they have not already done so. It just so happens that the compiler I am using to test the string module does the sensible thing and allows terms like SIZE(dummy_arg) in spec-expr in elemental procedures as an obvious extension. It still makes it difficult for the SPLIT routine to be made elemental in the standard though. -- Lawrie Schonfelder Director, Computing Services Dept. The University of Liverpool, UK, L69 7ZF Phone: 44(151)794 3716, Fax: 44(151)794 3759