Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!unmvax!unm-la!lanl!cmcl2!yale!wald-david
From: wald-david@CS.YALE.EDU (david wald)
Newsgroups: comp.lang.c
Subject: Pure Functions (was Re: C optimizer)
Message-ID: <50915@yale-celray.yale.UUCP>
Date: 16 Feb 89 21:39:17 GMT
References: <1398@quanta.eng.ohio-state.edu> <2008@goofy.megatest.UUCP>
Sender: root@yale.UUCP
Reply-To: wald-david@CS.YALE.EDU (david wald)
Organization: Yale University Computer Science Dept, New Haven CT  06520-2158
Lines: 25

In article <2008@goofy.megatest.UUCP> djones@megatest.UUCP (Dave Jones) writes:
>From article <1398@quanta.eng.ohio-state.edu>, by rob@kaa.eng.ohio-state.edu (Rob Carriere):
>> What seems to be missing is the idea that sleep *does* modify
>> something, namely time.  So formally speaking, your compiler should
>> consider sleep to have a side effect on a variable called __time.  If
>> you do that, there's no problem.
>
>Huh??
>
>On a time-shared system (such as Unix), does every instruction have
>a side effect on a variable called __time?

Point taken, but we still need a way to describe the difficulty with
treating sleep() as a pure function.  The actual difficulty is related
to the problem with volatile variables.  The compiler can't optimize
sleep() calls away, since they rely on some external (volatile) clock
variable (or some system call accessing this variable) for their
behavior.


============================================================================
David Wald                                              wald-david@yale.UUCP
waldave@yalevm.bitnet                                 wald-david@cs.yale.edu
"A monk, a clone and a ferengi decide to go bowling together..."
============================================================================