Email List: Xaustin-group-lX
[All Lists]

RE: Teleconference Minutes from 16th December 1999

To: yyyyyyyyyyyyyyy@xxxxxx, yyyyyyyyyyyy@xxxxxxxxxxxxx, yyyyyy@xxxxxxxxxxxxx
Subject: RE: Teleconference Minutes from 16th December 1999
From: yyyyyyyyyyyyyyy@xxxxxx
Date: Thu, 30 Dec 1999 06:26:45 +0100 (MET)
    From yyyyyy@xxxxxxxxxxxxx Tue Dec 28 20:09:11 1999

    One of the other subtleties of definition writing is that the
    IEEE rules (quoted to this list some time before the meeting)
    prohibit the use of the term being defined in the definition.
    (The idea is that the definition can be textually substituted
    for the defined term and the original sentence still makes sense.)

    Thus, trying to "fix" JUST that:

    >     Child Process: a new process created by another process 
    > [using fork()
    >     or spawn()] is known as a child process.
    > 
    > ... is known as a child process of that process.

    Child Process: a new process created [by fork() or spawn()] by a given
    process.  A child process remains the child of the creating process as
    long as both processes continue to exist.

It seems to me that you are creating problems by thinking
that there is a noun phrase Child Process. Instead, there is
a relationship "being a child of", "being a parent of".

Thus, given the definition

[1] Child Process: a new process created by another process
    is known as a child process of that process.

the IEEE replacement works like this:

    Whenever it is stated that `a is a child of b', we may read
    `a is a new process created by b'.

In other words, what is defined is not Child Process, but

[2a] Child Process of x: a new process created by x.

or, attempting to turn mathematics into (ugly) English:

[2b] Child Process of a process: a new process created by that process.

One might consider removing the `new'.



Andries

[yes, I am a mathematician]

<Prev in Thread] Current Thread [Next in Thread>