Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site utcsri.UUCP
Path: utzoo!utcsri!greg
From: greg@utcsri.UUCP (Gregory Smith)
Newsgroups: net.lang.c
Subject: Re: CTRL(x)
Message-ID: <3552@utcsri.UUCP>
Date: Wed, 29-Oct-86 13:46:31 EST
Article-I.D.: utcsri.3552
Posted: Wed Oct 29 13:46:31 1986
Date-Received: Wed, 29-Oct-86 16:04:32 EST
References: <4880@brl-smoke.ARPA>
Reply-To: greg@utcsri.UUCP (Gregory Smith)
Organization: CSRI, University of Toronto
Lines: 28
Summary: 

In article <4880@brl-smoke.ARPA> kb5mu@NOSC.ARPA writes:
>>But. . .is there a way to write an ANSI cpp version of
>>        #define CTRL(c)
>
>I don't have an ANSI cpp handy (does anybody?), so I can't test
>this, but how about using the ANSI invention of # for "string-izing":
>        #define CTRL(c) (#c[0] & 037)
>This should turn
>        CTRL(z)
>into
>        ("z"[0] & 037)
>which would get what you wanted, except that relatively stupid
>compilers might allocate storage for the string literal "z".

It's worse than that. "z"[0] is *("z"), and unary * is not allowed
in constant expressions. So ALL compilers would allow storage for "z"
and then generate code to read that storage at run-time.
Of course, ``case CTRL(c):'' wouldn't work either.

You don't expect C compilers to pull constants out of string literals,
do you?
[ I don't have an Ansi draft, so as far as I know there may be a rule
saying that constant expressions may contain string literals indexed by
constants. But I doubt it].

-- 
----------------------------------------------------------------------
Greg Smith     University of Toronto      UUCP: ..utzoo!utcsri!greg