Project 34: Transportability tutorial Harwell 1980: ============= Ford presented his first thoughts on a transportability tutorial (see IFIP/WG 2.5 (Harwell-34)734) and invited comments. Lawson queried whether run-time efficiency and programmer effort were really just as good under subset Fortran. Gear stressed the need for certain heuristic parameters in portable software. It was agreed that a sub-committee consisting of Ford (Chairman), Smith and Fosdick should prepare a full tutorial on the basis of Ford's work, for presentation at the next meeting. Ford stressed that he wanted to include mention of explicit problems on explicit machines. The need for examples in the paper was agreed. Madison 1982: ============= Ford wondered whether there was still a need for a transportability tutorial. The group felt strongly that wide dissemination of the ideas was very important though they were no longer new. Ford agreed to prepare a draft for circulation to the group before the end of the year. It was agreed that the document should be edited by Ford, Fosdick and Smith, following the agreed publication policy (IFIP/WG 2.5(Harwell-36)736) without official IFIP approval. Soederkoeping 1983: =================== Ford had promised to write a draft on the matter. Vouk, Einarsson volunteered to help. Ford agreed to send a shorter paper by the end of 1983. Pasadena 1984: ============== Transportability tutorial (Ford, Feldman, Gentleman) ---------------------------------------------------- Ford initiates the discussion on a transportability tutorial. He notes that in the early 70s the problem of carrying FORTRAN programs between different computing environments was generally recognised as a problem for the scientific and technical computing community that urgently required a solution. Through specification of a FORTRAN intersection subset and parameterization of a model environment, a methodology of algorithm design and software development evolved which, with compromises between tradeoffs, effectively overcame the problem. The introduction of FORTRAN 77 has been charging the situation (with its need for a new language subset). Further, the general introduction into the marketplace of early scientific workstations has demonstrated the importance of graphical standards, workstation characteristics interactive interrupters and a number of other features that need to be included in our model of a computing environment. The enormity of the model revision becomes apparent if one considers the terminals/workstations - mini-main frame /supercomputer/special architecture - programming environments/application packages/ graphical systems/data bases - networks for which a program is conceived and within which it is developed, used, maintained and The transportability tutorial would cover the recognition of the problem in the early 70 s ; why the transportability approach met the needs of a number of different types of users - involving a number of different forms of numerical software; and the places where this approach was still valuable. However it would also discuss the need for a profound review of the model due to the evolving technology in language, arithmetic, graphics, software and hardware. Feldman now discusses this in more detail. The new world offers a wide spectrum of choices and opportunities. We have not reached a stage where we understand the overall range of relevant possibilities, but it is clear that the change in the marketplace and the technology makes consideration of a wider field of portability very attractive. a) Output is no longer adequately described by the line printer model: color graphics is becoming common b) Input is no longer lines of characters - pointing devices and voice input are available. c) Computing engines are no longer sequential uniprocessors: expensive co-processor chips and multiprocessor computers are now available. d) Program environments vary widely and subtly; the users are less competent or willing to expend great efforts to accommodate limitation of supplied software. We really need to extend our paradigm to encompass the different dimensions. --------------------- Feldman, Ford, Gentleman will write a paper to be published in CACM. Smith has a pessimistic view of the difficulties of creating software in an environment where no model exist. It is felt that we are in a transition period, hence the range of variations offered will focus down to only 2 or 3 in 5 years from now. Gentleman: the paper should not discourage people. Ford: it should focus on the issues. Lawson: it should give examples of reasons for portability, and classes of software. Ford: a methodology based on compromise. Wright: a practical paper. Lawson: portable versus reusable software. A draft version of the paper will be sent to Aird, Cody, Einarsson, Lawson, Reid, Vouk, Wright. Any other member who wants the paper writes to Ford. Sophia-Antipolis 1985: ====================== Transportability (Feldman) -------------------------- A very early draft of a paper was distributed, and comments solicited. An improved draft will be circulated. The basic theme is that the consensus that portability was reasonably understood has broken down. The older language-based model is still valid, but the range of problems that people feel ought to be solved has increased astonishingly. The hardware spectrum ranges from tiny and cheap to super and expensive, from high-resolution colour to teletype output, from heavily supported central installations to personal machines owned by non-experts. The desire to move complicated programs makes it difficult to restrict the amount of information that must be transported along with the obvious code. To regain mastery of the portability question, we must find suitable places to cut the problem, and interfaces upon which we can agree. After a discussion reflecting the difficulty of the problem, the Group agreed that a paper should be written, as positive as possible, which delimits the possibilities. Feldman will send a first draft to the Group and members of project 34 (see document IFIP/WG 2.5 (Sophia-Antipolis-10) 1210). Como 1987: ========== Document: IFIP/WG 2.5 (Como-10) 1410, 12 pages. A discussion developed around the paper presented by Feldman. It was noted that the tone of the paper was somewhat too pessimistic, and that although it is true that portability is still a problem and that the claims on portability found in the literature need balancing, the paper should offer a less pessimistic tone in its published version. (Feldman, Lawson, Delves, Ford, Rice, Vouk, Gentleman) Stanford 1988: ============== Document: IFIP/WG 2.5 (Como-10) 1410, 12 pages. The following is the summary of the report presented by S. Feldman . The authors (Feldman and Gentleman) plan to update and improve the manuscript distributed at the last meeting (1410) and the submit it for publication. He noted that completed one part of the project, but that another presentation is planned the next WG 2.5 meeting. Jerusalem 1990: =============== Documents: IFIP/WG 2.5 (Como-12) 1410, 12 pages. Feldman informed the group that a paper he has written with Gentleman (Como meeting document 1410) will be published. He suggested that the project should be closed, and language independent arithmetic should be considered in project [63]. Document: ========= "Controversy: Portability - A No Longer Solved Problem" by Stuart Feldman and W. Morven Gentleman, Computing Systems Journal (the journal of the Usenix Association) volume 3, number 2 (Spring 1990), pp. 359-380.