Usenet 1994 October/usenetsourcesnewsgroupsinfomagicoctober1994disk2.iso/std_unix/volume.30/text0095.txt

Text File | 1993-03-11 | 5.2 KB | 121 lines

Submitted-by: [e-mail] (Jeffrey S. Haemer)

           USENIX Standards Watchdog Committee

       Stephen R. Walli <[e-mail]>, Report    Editor


       POSIX.18: POSIX Platform    Environment Profile


       Paul Borman <[e-mail]> reports on the January 11-15,
       1993 meeting in New Orleans, La.:

       The title of POSIX.18 reads:

       ``Draft Standard    for Information    Technology -- POSIX
       Standardized Profile -- USI-P001    POSIX Platform.''

       The title says it all.  What is a POSIX Standardized
       Profile?     What does USI-P001 mean?  If you can answer these
       questions - We Want You!     Unfortunately,    this confusion is
       carried through the whole draft.

       Due to problems of redundancy and obfuscation, the working
       group unmercifully hacked away at the draft with    an axe in
       the previous meeting.  This meeting we took out our Bowie
       knives to whittle it down further still.

       The three major issues discussed    were the prose,    what was it
       for, and    what new normative text    or changes to the normative
       text should be made.

       This discussion of the prose centered around the    large
       amounts of redundant and    apparently meaningless text in the
       draft.  Was it boiler plate?  Was it just the previous
       editor?    Did it simply come from    Planet X in the    middle of
       the night?  Extra-terrestrial or    not, either the    prose was
       simply removed or we reworded it    to be more easily
       understood.

       First on    the list of things to be clarified was the
       introduction, which was determined to be    mostly redundant or
       irrelevant.  We did decide to reword it to indicate that
       POSIX.18    describes UNIXTm] Classic or Version 7,    for those
       who remember it.     The profile still will    not define
       administrative interfaces, or even a way    to login.

       We did lobby the    POSIX.1a working group to modify a couple
       of interfaces to    bring them in line with    FIPS 151-2.  [Ed -
       FIPS 151-2 is the updated NIST specification of the POSIX.1
       standard, used in U.S. government procurements where POSIX-
       like functionality is required.]     We hope POSIX.18 will
       mirror this new FIPS.  These modifications were:

      - When read()    or write() is interrupted by a signal,
        after having read/written any data,    then they will
        return the byte count instead of -1,

      - that the group-ID of a file    at creation time is that of
        the    directory in which it is created, and not the
        effective group-ID of the process.

       We introduced text in POSIX.18 that requires that CS7, CS8,
       CSTOPB, PARODD and PARENB be supported from the POSIX.1
       General Terminal    Interfaces section.  We    are not    clear
       exactly what NIST was trying to accomplish by this and any
       comments    would be appreciated.

       There were several parts    of the document    that existed to
       fulfill TR10000 requirements, but as TR10000 is changing,
       much of this was    removed.  [Ed. - TR10000 is the    ISO
       technical report, defined originally in the OSI profile
       world, and now making itself felt in the    POSIX profile
       space.]    We are going to    lobby for the new TR10000 to
       require less and    make it    easier to understand.

       We restructured the two or three    pages of real normative
       text in the document in line with our decision in the last
       meeting to require the C    language.

       Due to a    new SEC    ruling,    we plan    to remove the current,
       inadequate test assertions in the document, and concentrate
       on the normative    text.

       Our major additions to the normative text, aside    from the
       FIPS 151-2 item mentioned earlier, were coherency
       statements.  We have required, for example, that    all the
       base standards that are pointed to by this profile must be
       implemented with    the the    same file-system name space and    use
       a consistent byte size.

       We also mandated    that text files    would be usable    between    all
       the different base standards and    that text files    can be used
       to contain source code that the compilers can compile.
       Without these sorts of statements it would have been
       technically possible to have a conforming system    in which vi
       was not capable of creating a file that the C compiler could
       compile!

       Other things were that the shell    could execute a    program
       built with the compilers    and that the compiler would allow
       use of the POSIX.1 functions.  Pretty straightforward and
       obvious stuff, but that is the sort of thing a profile must
       point out to make itself    useful.

       Overall I feel that the POSIX.18    draft made a lot of forward
       progress, but because it    now references POSIX.1a    it cannot
       go to ballot.  We also feel we need to do a bit more work
       cleaning    up the wording of the draft (and come to grips with
       what NIST is really asking in FIPS 151-2).

       Please note that    POSIX.18 is the    profile    that will more than
       likely define the basics    of a timesharing UN*X system in    the
       future.    If you are concerned about this    you might want to
       show up at our next meeting, and    you will certainly want    to
       join the    balloting group.


Volume-Number: Volume 30, Number 96