Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site bbncca.ARPA
Path: utzoo!linus!decvax!bbncca!jsol
From: jsol@bbncca.ARPA (Jon Solomon)
Newsgroups: net.ai
Subject: Re: Halting Problem: Resource Use
Message-ID: <364@bbncca.ARPA>
Date: Thu, 1-Dec-83 20:35:15 EST
Article-I.D.: bbncca.364
Posted: Thu Dec  1 20:35:15 1983
Date-Received: Fri, 2-Dec-83 08:15:16 EST
References: <13223@sri-arpa.UUCP>
Organization: Bolt, Beranek and Newman, Cambridge, Ma.
Lines: 23


I was looking at the infinite loop problem, and propose that if you look at
it as a timesharing problem thusly:

	A user writes a program which is stuck in an infinite loop. He
doesn't know how to get out of it, or even that it exists. He just runs his
program (I'm thinking of the old days when we ran programs on cards). Sooner
or later someone else (an operator or systems programmer in this case) sees
the job chewing up resources and not doing anything useful. He kills the job
and sends a note (if he can) or otherwise gets in touch with the user
telling him that something needs to be looked at. The key here is that one
need not build a foolproof mechanism for "self awareness" or "self
observation" as long as there is some other entity who can look at the
state and interrupt its context if he thinks you need service.

Some people are in infinite loops for their entire lives (opinion).
Their "service interrupts" are ignored.
-- 
[--JSol--]

JSol@Usc-Eclc/JSol@Bbncca (Arpa)
JSol@Usc-Eclb/JSol@Bnl (Milnet)
{decvax, wjh12, linus}!bbncca!jsol