Review of THE

From: Greg Green (ggreen_at_cs.washington.edu)
Date: Tue Jan 06 2004 - 19:42:20 PST

  • Next message: Steve Arnold: "Steve Arnold: Review of Dijkstra, "THE""

    This paper describes an early operating system. The system is not
    multiuser but does run multiple processes. The hardware had drum
    storage, core memory, and used interrupts. It also had several
    peripherals. The os used a concept of paged memory, with the allowed
    memory space being much larger than the available memory, with pages
    being swapped to the drum storage when necessary. The os was designed
    with 5 different logical levels of abstraction. Each level provided
    abstraction for the layer above it. Level 0 was scheduling and timer
    interrupt management. Level 1 controlled the drum, the management of
    pages, and associated interrupts. Level 2 had messages to and from
    peripherals. And so on.

    The paper made some questionable assertions about what an outstanding
    and bug-free system they had made. It appears that the review
    committee had the same opinion and asked for some backup, giving us
    appendix a. As an aside, I found his comments on how having people
    working on something half-time was very detrimental to productivity. I
    have also found this to be true.

    He also talked at some length about having thorough testing programs
    at each level of abstraction. He also mentioned that these types of
    tests are difficult to conceive and implement. I would agree
    whole-heartedly. That is probably why I see it happen so rarely in
    practice, therefore, the low level of quality in a lot of software.

    On the whole, the paper is interesting to see how many of the concepts
    of a modern operating system were implemented such a long time ago (in
    terms of computer science measure)


  • Next message: Steve Arnold: "Steve Arnold: Review of Dijkstra, "THE""

    This archive was generated by hypermail 2.1.6 : Tue Jan 06 2004 - 19:42:22 PST