From jls@uxb.liv.ac.uk Mon Feb 1 13:43:15 1993 Received: from ib.rl.ac.uk by dkuug.dk with SMTP id AA15948 (5.65c8/IDA-1.4.4j for ); Mon, 1 Feb 1993 14:50:35 +0100 Received: from mail.liv.ac.uk by ib.rl.ac.uk (IBM VM SMTP V2R1) with TCP; Mon, 01 Feb 93 13:43:24 GMT Received: from uxb.liverpool.ac.uk by mailhub.liverpool.ac.uk with SMTP (PP) id <03879-0@mailhub.liverpool.ac.uk>; Mon, 1 Feb 1993 13:43:24 +0000 From: "Dr.J.L.Schonfelder " Received: from uxb.liv.ac.uk (uxf.liv.ac.uk) by uxb.liv.ac.uk; Mon, 1 Feb 93 13:43:16 GMT Message-Id: <18595.9302011343@uxb.liv.ac.uk > Subject: HPF 1.0 comments Chapter 2 To: hpff-comments@edu.rice.cs Date: Mon, 1 Feb 93 13:43:15 GMT Cc: SC22WG5@dkuug.dk (SC22/WG5 members), UKFORTRAN@edinburgh.ac.uk (UK Fortran Panel) X-Mailer: ELM [version 2.3 PL11] X-Charset: ASCII X-Char-Esc: 29 Here are my comments on Chapter 2 Comment on HPF 1.0, Jan'93 Chapter 2, Terms and Concepts I heartily approve of the choice of Fortran 90 as the base from which to build HPF. I also approve in general recommending restrictions on the use of Fortran 90 storage association facilities. As a teacher of Fortran 90 to research users and students alike, these facilities are only covered in my courses as examples of obsolete archaisms to be avoided in new code but which will be found in old Fortran 77 codes. I also think the general approach to extending Fortran 90 by directives and making only limited extensions to the basic syntax of the language is sound. However I think the use of directives may have gone too far. A guiding principle I would have thought would be that a directive is appropriate where the effect of the directive is entirely lacking in other linguistic significance. All the directive does is to provide information that may be used by the processor to help optimisiation but otherwise has no other semantic or syntactic effect on the resulting code or the results of executing it; it merely may run faster. Where the presence of a qualifier has an effect on the contexts that an entity may be used, such as the PURE designation, or where the code covered by the qualifier is restricted so that the action of the processor may produce erroneous results if the restrictions are not obeyed, then it would be better if these were language syntax rather than directives. This I think would suggest the memory alignment and distribution extensions are rightly provided by directives that can be ignored or not without otherwise effecting the program. Even if HPF extensions were to be adopted by the formal language standard there are strong arguments for retaining the comment/directive character of such facilities. However, the PURE and INDEPENDENT qualifiers, for example, I think would be better handled as additional syntax. This will be discussed more fully in subsequent comments covering the specific chapters. -- Dr.J.L.Schonfelder Director, Computing Services Dept. University of Liverpool, UK Phone: +44(51)794 3716 FAX : +44(51)794 3759 email: jls@liv.ac.uk