From J.Reid@directory.rl.ac.uk Mon Dec 14 20:03:28 1992 Received: from danpost4.uni-c.dk by dkuug.dk with SMTP id AA04799 (5.65c8/IDA-1.4.4j for ); Mon, 14 Dec 1992 19:02:35 +0100 X400-Received: by mta danpost4.uni-c.dk in /PRMD=minerva/ADMD=dk400/C=dk/; Relayed; Mon, 14 Dec 1992 19:01:59 +0100 X400-Received: by /PRMD=uk.ac/ADMD= /C=gb/; Relayed; Mon, 14 Dec 1992 19:01:53 +0100 X400-Received: by /PRMD=UK.AC/ADMD= /C=GB/; Relayed; Mon, 14 Dec 1992 18:49:15 +0100 X400-Received: by /PRMD=UK.AC/ADMD= /C=GB/; Relayed; Mon, 14 Dec 1992 19:03:28 +0100 X400-Received: by /PRMD=UK.AC/ADMD= /C=GB/; Relayed; Mon, 14 Dec 1992 19:03:28 +0100 Date: Mon, 14 Dec 1992 19:03:28 +0100 X400-Originator: J.Reid@directory.rutherford.ac.uk X400-Recipients: SC22WG5@dkuug.dk X400-Mts-Identifier: [/PRMD=UK.AC/ADMD= /C=GB/;<9212141803.AA01388@numerical.cc] X400-Content-Type: P2-1984 (2) Content-Identifier: Comments on t... From: " (John Reid)" Sender: J.Reid@directory.rl.ac.uk Message-Id: <9212141803.AA01388@numerical.cc.rl.ac.uk> To: baw@seg.npl.co.uk Cc: SC22WG5@dkuug.dk, payne@cra.enet.dec.com Subject: Comments on the Fortran in LIA-1 X-Charset: ASCII X-Char-Esc: 29 Brian, As promised when we met last week, here are my comments on the Fortran in LIA-1. I am copying this to WG5 and X3J3, and will let you know if any of my comments turn out to be controversial. Regards, John. .......................................................................... Comments on the Fortran in Draft International Standard ISO/IEC CD 10967-1:1992 LANGUAGE INDEPENDENT ARITHMETIC (LIA) - Part 1 John Reid, Atlas Centre, Rutherford Appleton Laboratory, Didcot, Oxon OX11 0QX England email: jkr@directory.rl.ac.uk Annex B.5 FORTRAN 77 Either this section should be removed or it should be rewritten so that a Fortran 77 / LIA-1 program is also a Fortran 90 / LIA-1 program. Upwards compatibility has always been a very important design goal for Fortran 90, and I think that the same should apply here. A user should be able to migrate an application from Fortran 77 to Fortran 90 with little or no change. My preference is to remove this section, since Fortran 77 is no longer an international standard. Annex B.6 Fortan 90 Fortran 90 does not permit reserved words. The effects required can, however, be obtained by packaging LIA-1 as a module, say called LIA_1. To access the LIA-1 facilities, a user would need need to include a USE statement for LIA_1. Note that the USE statement permits the local renaming of module entities. The type and kind type parameter of every procedure and constant of the module (i.e. all those labelled with a dagger) needs to be specified. Fortran 90 regards underscore as alphanumeric and allows names to have up to 31 characters so the LIA names that contain underscores, such as 'rnd_error' and 'set_indicators', can be used as Fortran 90 names. It would be more user-friendly to make RNDSTYLE (renamed rnd_style) have type character*8 with allowed values 'nearest', 'truncate', 'other', with the case being ignored. MODULO is a Fortran 90 intrinsic, so no dagger is needed. The alternative syntax for '.NE.' is '/=', rather than '<>'. Annex C.3 Delete 'inquiry' from the second line of both tables since the new functions are not inquiry functions. Annex C.4 The directive should make it obvious that this is an LIA directive. I suggest !LIA$ directive to conform to the lead being set by HPF: !HPF$ directive