Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!haven!adm!smoke!gwyn
From: gwyn@smoke.BRL.MIL (Doug Gwyn )
Newsgroups: comp.lang.c
Subject: Re: programming puzzle (silly)
Message-ID: <9846@smoke.BRL.MIL>
Date: 14 Mar 89 01:56:19 GMT
References: <24820@amdcad.AMD.COM> <218@umigw.MIAMI.EDU>
Reply-To: gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>)
Organization: Ballistic Research Lab (BRL), APG, MD.
Lines: 21

In article <218@umigw.MIAMI.EDU> steve@umigw.miami.edu (steve emmerson) writes:
>In article <24820@amdcad.AMD.COM> tim@amd.com (Tim Olson) opines that the
>expression "n&&m*=n--" is incorrect C:
>I belive the correctness of a expression is determined by whether or not
>the expression can be *generated from* the grammer's production rules; not
>whether or not a given implementation of a compiler can parse it.

Right.

>The given expression is correct C as it can be generated by the following
>rules:

I didn't much like your particular production.  Using the pANS grammar,
the conclusion that the parse is NOT effectively "(n&&m) *= n--" is a
correct deduction; the key point is that "n&&m" cannot possibly be parsed
as anything that is allowed as the left operand of the assignment
expression (called a "unary expression" in the pANS grammar).  However,
the grammar does not support the production you gave, either.  At least
one set of parentheses would have to be added to obtain a valid
expression.  The phrase-structure grammar does not in general allow
arbitrary expressions as operands, only limited classes of expressions.