From adt10@amdahl.uts.amdahl.com Tue Mar 17 02:59:17 1992 Received: from mcsun.EU.net by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8) id AA07021; Tue, 17 Mar 92 02:59:17 +0100 Received: from CHARON.AMDAHL.COM by mcsun.EU.net with SMTP id AA00449 (5.65a/CWI-2.153); Tue, 17 Mar 1992 02:59:15 +0100 Received: from amdahl.uts.amdahl.com by charon.amdahl.com (4.0/SMI-4.1/DNS) id AA15787; Mon, 16 Mar 92 17:57:55 PST Received: by amdahl.uts.amdahl.com (/\../\ Smail3.1.14.4 #14.9) id ; Mon, 16 Mar 92 17:59 PST Message-Id: Date: Mon, 16 Mar 92 17:59 PST From: adt10@uts.amdahl.com (Andrew D. Tait) To: sc22wg5@dkuug.dk Subject: WG5L12 Proposal - The Wrong Approach X-Charset: ASCII X-Char-Esc: 29 I am both concerned and disappointed that the WG5L12 committee has developed a proposal for the future management of Fortran Standards activities which merely embodies the practices and procedures that have failed us so miserably in recent years. I understand that there is pressure from the convenor to complete this exercise by the end of March. Attached are the comments that I have already sent to X3J3 and I sincerely hope that those of you who feel as strongly as I do that the proposed approach is wrong will give your representatives on the WG5L12 committee the appropriate instructions to oppose the plan. The proposal is much too close to what we have done in the past. The words (of which there are too many) are different, but it is easy to see that it can easily degenerate into a mess worse than that we experienced over the last few years. WE NEED A RADICALLY DIFFERENT APPROACH. What follows are my initial ideas and I have not got all the details fleshed out. We must look at the task to be accomplished and assign and manage our resources with that in mind. Whether the resulting structure fits with the organization of WG5 and the various national bodies is irrelevant at this time. Schedule ________ This must be PRECISE, no wishy-washy nonsense of "about the year 1995", or "about the year 2000": 1. We will produce a maintenance revision in 1995. 2. We will produce a complete revision by 2000. 3. We will only undertake work which can be completed within this schedule. 4. We will organize and manage our resources to achieve these objectives. Organization ___________ WG5 meeting once a year can be no more than a review and approvals body. It cannot effectively manage anything. I propose the following: 1. Requirements/Architecture group. Develops the requirements document and gets WG5 approval. Manages the development groups to ensure that what they are doing is compatible and will fit togther properly. This group has editorial responsibility for all drafts and the final standards document. The crucial point is that this group should be limited to 5 or 6 people who are willing and able to make a commitment to do this work. This means that they must have long-term financial backing. Input to this group can come from any source. 2. Development groups. If we are to effectively use the available resources and to meet schedules there must be more than one development group. These must also be small committed groups. A number of small development groups working on components with clearly defined rules for integration will make faster progress than a monolithic organization. The work plans for these groups will be defined by the Requirements/Architecture group. This will go a long way to ensuring that creeping fucntionality is avoided. It also allows for the same assignment to be given to more than one group to provide alternative solutions to the same problem. No other groups or organizations will be allowed to do development work which will be considered for inclusion in the standard. This means that if the architecture group discovers that components do not fit together, then the groups responsible for the offending parts must work together to fix the problem. We must put more trust in these small groups than we have done in the past and this approach should help prevent the national bodies or WG5 from a wholesale redesign of various parts of the language. Content _______ The maintenance revision will be limited to corrections, clarifications, and interpretations. NO CREEPING FUNCTIONALITY. The complete revision will be limited to those features in the approved requirements document. The requirements document will be frozen in 1997 and at that time the final content of the complete revision will be specified with due regard to whether there is actually enough time to complete the development work. Obviously "collateral standard" is the in thing at the moment. I am vehemently opposed to collateral (i.e. second class) standards. Let us be clear about the reason why we got ourselves into this silly situation in the first place. DIN wanted the varying string feature in Fortran 90. Nobody else on WG5 supported this feature, but because DIN insisted on voting NO if they did not get something, WG5 let them go ahead with the collateral standard. Forget the rationale that has been produced for the Vary String collateral standard there was (and probably is) no demand for this feature in the international community. At the Lund WG5 meeting there were a number of eloquent speeches against collateral standards. The bottom line is that it is not reasonable to shackle the requirements or development groups to a commitment to include any collateral standards, they may just not fit into the overall architecture of the revised standard. New collateral standards will NOT be produced as part of the effort to revise Fortran 90. Representation of, and Access to the Standard _____________________________________________ The standard will be available in machine readable format and no organization, national or international, shall have the right to copyright it. The document will be available to all on request for, at most, a fee to cover the cost of reproduction, media, and shipping. A master must be identified and protected against changes. Reviews _______ The rules for conducting reviews of proposed standards must be tightened. I recommend that we adopt a set of rules modeled on those of IEEE. Obviously we probably cannot require people to pay fees for the privilege of reviewing proposals, but we can insist on specific formats and require comments to be submitted in machine readable form SO THAT WE CAN GET OUR JOB DONE ACCORDING TO THE SCHEDULE. Comment _______ If we have to change the way X3 and/or ISO does business in order to make more effective use of the resources available, we must not hestitate to do so. Andrew D. Tait **************************************************************************