Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!unmvax!pprg.unm.edu!hc!lll-winken!uunet!tektronix!tekcrl!tekchips!stevev
From: stevev@tekchips.LABS.TEK.COM (Steve Vegdahl)
Newsgroups: comp.lang.misc
Subject: Re: Language Design
Summary: let's stop picking nits
Message-ID: <3746@tekcrl.LABS.TEK.COM>
Date: 23 Mar 89 19:24:57 GMT
References: <5200040@m.cs.uiuc.edu> <12443@watdragon.waterloo.edu> <9122@claris.com>
Sender: ftp@tekcrl.LABS.TEK.COM
Lines: 23

In article <9122@claris.com>, hearn@claris.com (Bob Hearn) writes:
> Enough of this bullshit!  How many times do I have to say it???  The grammar
> in K&R is context-free.
 ...
> Now, before anyone else wants to tell me that K&R's grammar for C is not
> context-free, will you *please* go and look up the definition of context-free
> grammar???  Hint: anything that can be expressed in BNF is context-free!!

I detect that we are picking nits here.  I think that we all know what is
meant by the claim that the K&R grammar is not context-free.  It has been
noted that the K&R grammar is *ambiguous*; hence, its specification of
how programs should be parsed is *ill-defined*.  Additional
"context-sensitive" specifications are added to the K&R grammar in the form
of English prose (in the second edition, anyway; the first edition
seems to ignore the ambiguity problem).

Effectively, the grammar in K&R is "BNF + English prose".  Taking this
view, the K&R grammar is not context-free.

		Steve Vegdahl
		Computer Research Lab
		Tektronix Labs
		Beaverton, Oregon