Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83 (MC830713); site tjalk.UUCP
Path: utzoo!linus!decvax!mcvax!vu44!tjalk!dick
From: dick@tjalk.UUCP (Dick Grune)
Newsgroups: net.lang.c
Subject: Re: a ? b = c : d, and what the Manual says
Message-ID: <307@tjalk.UUCP>
Date: Wed, 21-Mar-84 08:09:27 EST
Article-I.D.: tjalk.307
Posted: Wed Mar 21 08:09:27 1984
Date-Received: Sat, 24-Mar-84 16:34:27 EST
Organization: VU Informatica, Amsterdam
Lines: 43

Since here in Europe the net news often arrives a bit late, I generally do
not bother to answer, because anything I may have to contribute
usually appears repeated N times in the next few rounds of news.

Nevertheless, in the protracted discussion about the acceptability of

	a == b ? c = d : e = f

I have not to date seen the essential remark that the reference
manual answers the question, so here I go.

I quote (RM 7, first paragraph):
	"The precedence of ... operators is the same as the order of
the major subsections... . Thus, for example, the expressions
referred to as the operands of + ($7.4) are those defined in
$$7.1-7.3."

This effectively says how to expand the grammar in the manual to
obtain a more complete one. For $7.13 this expansion results in:

conditional_expression:
	expression7_13 ? expression7_13 : expression7_13
expression7_13:
	logical_or_expression	/* 7.12 */
	logical_and_expression	/* 7.11 */
	...
	unary_expression	/* 7.2 */
	primary_expression	/* 7.1 */

This grammar does not produce

	a == b ? c = d : e = f

and you may draw your conclusions from that.

If this has been on the net, the news just hasn't come to me, and I
apologize for wasting time and bandwidth.

					Dick Grune
					Vrije Universiteit
					Amsterdam

And my name isn't Richard!