NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/lang/c/12478

Internet Message Format | 1992-08-18 | 3.5 KB

Path: sparky!uunet!sun-barr!ames!agate!stanford.edu!rutgers!concert!sas!mozart.unx.sas.com!sasghm
From: [e-mail] (Gary Merrill)
Newsgroups: comp.lang.c
Subject: Re: Pointer/address reluctance
Message-ID: <[e-mail]>
Date: 18 Aug 92 13:01:38 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (Noter of Newsworthy Events)
Organization: SAS Institute Inc.
Lines: 54
Originator: [e-mail]
Nntp-Posting-Host: theseus.unx.sas.com


In article <[e-mail]>, [e-mail] (Peter van der Linden) writes:
|> From: [e-mail] (Systems Anarchist)
|> > You are perfectly free to redefine 'address' to mean 'pointer.'
|>
|> It's not a redefinition, as it is already used in the ANSI C standard
|> with this meaning.
|>
|> There may well be archaic or unorthogonal hardware that defines "address"
|> in some special, different, non-intuitive manner.   The rest of computer
|> science has known for some time that large, flat, byte-addressable address
|> spaces are the way to go.
|>
|> This is not to criticize Steve who has clearly thought deeply about
|> the matter, and come up with a paradigm that works for him.  To me,
|> the effort to avoid equating "address" and "pointer" is somewhat like
|> trying to be more Catholic than the Pope.  If the ANSI C std doesn't
|> do it, why should anyone else?

I really must protest one last time this view that the ANSI C standard
equates "address" and "pointer".  I you must persist in this view, that
certainly is your right, but (as has been amply demonstrated by
several postings) there is in fact nothing in the standard that asserts
or implies such an equation.  Most programmers lead perfectly happy
lives without drawing the careful distinctions among syntax, semantics,
and implementation that language designers and compiler writers must
make.  Most programmers have much more limited goals:  get it to work
(or at least get it to *appear* to work most of the time), or get it
to work on this one machine (or on this one *type* of machine -- e.g.,
having a flat address space).  If identifying pointers and addresses
serves your purpose, then by all means do this.  But in that case it
is better (certainly more accurate) to say "Well, as a practical
matter I think of pointers as addresses," rather than insisting that
"The standard says that pointers and addresses are the same."

You can actually do a lot of practical mathematics without bothering
to distinguish numbers from the names of numbers (numeric constants),
or bothering to distinguish a vector from a letter with a little arrow
drawn over it, but most mathematicians eventually will have to make such
distinctions (if only, in the case of numbers,  because there are so
many more numbers than there are names for them).  Similarly, *some*
programmers will have to make the appropriate pointer/address distinction,
and others will benefit from making it.  The standard explicitly and
intentionally allows this.

In any event, your original question was for an explanation of why
the "experts" make such a distinction.  That has now been answered in
great detail from a number of quarters.  If you don't like the answer,
ignore it.

--
Gary H. Merrill  [Principal Systems Developer, C Compiler Development]
SAS Institute Inc. / SAS Campus Dr. / Cary, NC  27513 / [phone]
[e-mail] ... !mcnc!sas!sasghm