From JLS@liverpool.ac.uk Mon Aug 24 12:15:45 1992 Received: from vm.uni-c.dk by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8) id AA25350; Mon, 24 Aug 92 12:15:45 +0200 Message-Id: <9208241015.AA25350@dkuug.dk> Received: from vm.uni-c.dk by vm.uni-c.dk (IBM VM SMTP V2R2) with BSMTP id 6740; Mon, 24 Aug 92 12:16:29 DNT Received: from UKACRL.BITNET by vm.uni-c.dk (Mailer R2.07) with BSMTP id 6689; Mon, 24 Aug 92 12:16:28 DNT Received: from RL.IB by UKACRL.BITNET (Mailer R2.07) with BSMTP id 0085; Mon, 24 Aug 92 11:16:32 BST Received: from RL.IB by UK.AC.RL.IB (Mailer R2.07) with BSMTP id 2557; Mon, 24 Aug 92 11:16:04 BST Via: UK.AC.LIV.IBM; 24 AUG 92 11:16:00 BST Received: from UK.AC.LIVERPOOL by MAILER(4.4.t); 24 Aug 1992 11:12:18 BST Date: Mon, 24 Aug 92 10:48:14 BST From: Lawrie Schonfelder Subject: Re: Object orientation To: WEAVER@stlvm7.vnet.ibm.com, SC22/WG5 members In-Reply-To: Your message of Fri, 21 Aug 92 23:07:59 PDT X-Charset: ASCII X-Char-Esc: 29 On Fri, 21 Aug 92 23:07:59 PDT WEAVER@com.ibm.vnet.stlvm7 said: I agree with Dick. This is a key requirement, probably THE key requirement for F2000 or even F96. I think WG5 should have an OOP group not just J3. >Object oriented programming is getting significant >attention from the programming language community. > >It might be prudent for X3J3 to establish an ad hoc group to >study object orientation with the goals of: > > -- understanding object oriented programming in relationship to Fortran > (is Fortran 90 adequate? what kinds of extensions might be needed?) NO!!! F90 gets a long way down the line but experience with the string module shows a number of restrictions that stop proper semantic extension OOP programs from being developed. First and foremost derived types must be treatable as exactly paralleling intrinsic types, particularly where the internal structure of the type is PRIVATE. The user must be able to denote literal constants of the type, declare PARAMETER(CONSTANTS), Initialise variables, and perform proper I/O on objects of a derived type. Derived types must also be parameterised as are intrinsic types with KIND style parameters and LEN style parameters. There are other aspects of the general OOP concept that would be needed but fixing these defects in the current F90 definition would go a very long way to providing the necessary functionality in a Fortran like manner. > > -- understanding how an object oriented Fortran might relate to > C++ , COBOL, and possibly other programming languages. > > -- understanding the Object Management Groups (OMG) role, objectives, > and progress. > > -- determining if any Fortran products already provide "object orientation" > > -- recommending futher X3J3 action re object orientation, if any > >The results to be reported to X3J3 and, possibly, forwarded to WG5 as US input >to the requirements database. not possibly definitely! I think this should be a WG5 subgroup and I have no objection to it also reporting status of activity to J3 as well, as part of the essential coordination of requirement specification with the primary development body > >If such a group were to be established, I would suggest scheduling their final >report no latter than one year following establishment, with status reports at >each intervening X3J3 meeting. > >In establishing such a group, X3J3's first concern should be that the group have >an adequate skill mix. To achieve this, it will likely be necessary to locate >and include participants who are not members of X3J3. Yes! From WG5 and elsewhere >Thanks >Dick W Lawrie