Steve Arnold: Review of Ritchie, "UNIX"

From: Steve Arnold (steve.arnold4_at_verizon.net)
Date: Tue Jan 06 2004 - 21:19:54 PST

  • Next message: ahemavathy: "Review of Unix"

    Ritchie and Thompson of Bell Laboratories describe the UNIX operating system
    as it was implemented during the PDP-11/45 timeframe. They describe at
    length the file system, including how any files are stored (on disk) using
    indexes and a hierarchical directory structure. Much time is spent
    describing the file system, because almost everything is stored in it,
    including "special files" used for reading and writing I/O devices. They
    spend some time describing the sytem calls needed for using I/O, as well as
    a brief overview of the protection system. They then go on to describe in
    more detail how this is implemented logically and technically.
     
    In much briefer sections, they describe other aspects of UNIX. One of which
    is processes, noting that machine state is stored in images. It explains how
    the shell (interchangeable) can make system calls to basically command the
    operating system. Default syntax is given for such features as redirection
    and piping. In conclusion, they do a hand wave over performance and explain
    its success.
     
    It is quite interesting to me how much has not changed in UNIX. Most of what
    they explain in 1974 still holds true for UNIX-based operating systems
    today. They do describe some limitations (such as max file size) that are
    not realistic for today's computing needs. Given that it has withstood this
    30-year test of time, you would have though it would be more scalable from
    the beginning. Maybe this is somewhat of result of designing to the hardware
    available.
     
    In all, the paper was an excellent UNIX refresher for me, a paper that I
    might refer to one who would like some UNIX background. However, I do feel
    that they got wrapped up in some of the specifics and could have linked
    together some of the big features (such as explaining how processes and
    images are essential to multiprogramming). I would also liked to have know
    how their version differed from more primitive versions of UNIX and what
    have they learned.
     


  • Next message: ahemavathy: "Review of Unix"

    This archive was generated by hypermail 2.1.6 : Tue Jan 06 2004 - 21:19:23 PST