NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/lang/c/16600

Internet Message Format | 1992-11-16 | 1.7 KB

Path: sparky!uunet!olivea!spool.mu.edu!darwin.sura.net!gatech!concert!sas!mozart.unx.sas.com!sasghm
From: [e-mail] (Gary Merrill)
Newsgroups: comp.lang.c
Subject: Re: LHS Casting
Message-ID: <[e-mail]>
Date: 16 Nov 92 15:23:10 GMT
References: <[e-mail]>
Sender: [e-mail] (Noter of Newsworthy Events)
Organization: SAS Institute Inc.
Lines: 25
Originator: [e-mail]
Nntp-Posting-Host: theseus.unx.sas.com


In article <[e-mail]>, [e-mail] (D'Arcy J.M. Cain) writes:
|> I came across some of the ugliest code I have seen in a long time recently.
|> At one point there is a pointer to char that is declared as a pointer to
|> int.  To get an address within this block of bytes the programmer does:
|>
|> (int *) charptr = intptr + (OFFSET/2);
|>
|> Well of course this is bad.  The FAQ (question 2.13) shows another case

It's not only bad, it is a violation of ANSI.  See Section 3.3.4,
note 41 ("A cast does not yield an lvalue. ...")  Some older C
compilers used to accept this nonsense.  I cannot lay my hands on
a precise reference at the moment, but there was an issue of _Dr.
Dobbs_ some time in '85 or '86 in which a paper (to the ANSI C
committtee) written by Francis Lynch and me was reprinted almost in
its entirety as part of a column on the standardization of C.  In
that paper a number of compelling arguments are offered against
allowing a cast to be an lvalue.  If you can find the appropriate
issue of the magazine, you might be interested in these arguments.

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