Re: JavaMemoryModel: PC language alert: catatonia action

From: Eliot Moss (moss@cs.umass.edu)
Date: Mon May 17 2004 - 10:10:34 EDT


>>>>> "Bill" == Bill Pugh <pugh@cs.umd.edu> writes:

    Bill> OK, so names like spin, or busyWait aren't good names because they are
    Bill> behaviors that real programs do (e.g., busyWait on a volatile variable).

    Bill> Names like think, pause, stutter, and so on because they don't reflect
    Bill> the fact that the program will perform an infinite sequence of these
    Bill> actions.

    Bill> Some of the other names are just too cute.

    Bill> How about infiniteLoop action?

    Bill> Alternatively, blackHole?

Maybe this is not really an issue, but (except in certain formal semantics)
it troubles me to think of this infinitely busy behavior as a _single
action_. I think it is quite different from the other actions one considers
in the model, which (when they occur) consume finite (though not
specifically bounded) time (or perhaps can be viewed as occurring in an
instant). Jut expressing some uneasiness. It seems to be to be infinite
_repetition_ of a non-interacting (silent) action ...

Eliot
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:06 EDT