Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!elroy!aero!sm.unisys.com!ism780c!news
From: news@ism780c.isc.com (News system)
Newsgroups: comp.arch
Subject: Re: Semantics (was Software Distribution)
Message-ID: <17887@ism780c.isc.com>
Date: 13 Oct 88 00:25:10 GMT
References: <13889@mimsy.UUCP> <2472@uhccux.uhcc.hawaii.edu> <8916@ihlpb.ATT.COM>
Reply-To: marv@ism780.UUCP (Marvin Rubenstein)
Organization: Interactive Systems Corp., Santa Monica CA
Lines: 22

In article <8916@ihlpb.ATT.COM> nevin1@ihlpb.UUCP (55528-Liber,N.J.) writes:
>
>So, given that *NULL is an exception to dereferencing, what would be a
>good way to define it?
> _ __		NEVIN J. LIBER  ..!att!ihlpb!nevin1  (312) 979-4751  IH 4F-410

Might I suggest the definition used in the Pascal Standard.  First, the term
'error' means an error in the source program.  The standard says:  "It is an
error if the pointer-variable of an of an identified-variable denotes a
nil-value".  Or translated in to English, it is an error to dereference a
pointer to no object.

The standard also says: "A complying processor is required to provide
documentation concerning its treatment of errors".  So the way in which a
a processor treats an error is up to the implementer, the only requirement is
the implementor must inform the user.  A user who wants erroneous programs
to behave the same way on all implementations is just plain out of luck.

I hope the proposed C standard says approximately the same thing about
erroneous source programs.

    Marv Rubinstein