Steve Arnold: Review of Dijkstra, "THE"

From: Steve Arnold (steve.arnold4_at_verizon.net)
Date: Tue Jan 06 2004 - 20:55:20 PST

  • Next message: ahemavathy: "(no subject)"

    Although quite obviously an early paper, the Dijkstra paper tries to give an
    overview of "THE," and early multiprogramming system (not actually referred
    to as an operating system). He describes it in a progress report format,
    first giving an introduction to their project. He describes the concept,
    which is several layers of hierarchy, each mostly independent of one
    another. The layers are: 0) processor allocation, 1) segment controller, 2)
    console, 3) stream buffering, 4) user programs, and 5) operator.
     
    For a paper circa 1967, the author explains what are some fairly advanced OS
    concepts, albeit at a rather abstract level. The terminology has changed,
    but must of what is talked about exists in OSes today. (For example
    scheduling, referred to as processor allocation by the author.) It is
    interesting how he mentions that they did not allocate enough resources to
    debugging and the such, as this is what we spend a lot of time on when
    designing today.
     
    In my opinion, this is not one of Dijkstra's best works. His writing style
    was conversational and he spent a lot of time explaining why this paper
    would be not be too formal and why the project was not too structured. I
    also found that it was not very scientific. There was no data to support his
    results and no specific examples of how this was implemented. The setup of
    the paper, explaining the hierarchy, was the part that I found most
    valuable.
     


  • Next message: ahemavathy: "(no subject)"

    This archive was generated by hypermail 2.1.6 : Tue Jan 06 2004 - 20:54:48 PST