NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/lang/c/12290

Text File | 1992-08-13 | 2.7 KB | 57 lines

Newsgroups: comp.lang.c
Path: sparky!uunet!gatech!concert!sas!mozart.unx.sas.com!sasghm
From: [e-mail] (Gary Merrill)
Subject: Re: Pointer/address reluctance
Originator: [e-mail]
Sender: [e-mail] (Noter of Newsworthy Events)
Message-ID: <[e-mail]>
Date: Thu, 13 Aug 1992 17:28:22 GMT
References:  <[e-mail]>
Nntp-Posting-Host: theseus.unx.sas.com
Organization: SAS Institute Inc.
Lines: 43


In article <[e-mail]>, [e-mail] (Peter van der Linden) writes:
|> Can anyone explain the reluctance that some of the experts here
|> have in equating "pointer" with "address"?
|>
|> The ANSI C standard displays no such inhibition, e.g. Section 3.4
|> line 32 mentions "An address constant is a pointer to ..."
|>

First (and pardon what may appear as pedantry), the section you
quote does *not* "equate" a pointer with an address.  In fact, I
am unaware of anywhere the standard discusses the concept of *address*.
The section you quote (as well as section 3.3.3.2) establishes
a relation between a *pointer* and an *address constant*.  An
*address constant* is a kind of *constant* (it is *not* an address),
and a *pointer* is a type in C.  There is no type in C called
"address".  Thus, in part, the reluctance you sense may simply be
a resistance to "equating" a formal and precise concept (pointer)
with an informal concept (address) that is not given precise meaning
in the standard.

Beyond this, while there is a clear intuitive association of
"pointer" with "address", one may want to leave open the possibility
of implementing pointers as something other than addresses (perhaps
as an address together with some other component).  In theory,
I suppose one could conceive of an "abstract machine" and an
implementation of C for this machine that did not implement a pointer
as an address.  The semantics for C leaves this open (and should).
To do otherwise would be to limit the applicability of the language
by an arbitrary restriction on its semantics.

The issue here is conceptual and semantic.  To "equate" a pointer
with an address is akin to equating "array" with "contiguous sequence
of memory locations".  In short, it is to confuse a concept (or type)
with one *possible* implementation of that concept.  An array is
*usually* implemented as a contiguous sequence of memory locations,
but it need not be.  What characterizes an array is its "algebra" --
the set of operations that may be performed on it, the closure
conditions of such operations, etc.  Similarly for "pointer".
--
Gary H. Merrill  [Principal Systems Developer, C Compiler Development]
SAS Institute Inc. / SAS Campus Dr. / Cary, NC  27513 / [phone]
[e-mail] ... !mcnc!sas!sasghm