Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!pyramid!prls!philabs!linus!raybed2!rayssd!mirror!frog!john
From: john@frog.UUCP (John Woods)
Newsgroups: comp.lang.misc
Subject: Re: Language Design
Message-ID: <1163@frog.UUCP>
Date: 26 Mar 89 23:12:00 GMT
References: <5200040@m.cs.uiuc.edu> <12443@watdragon.waterloo.edu> <9165@claris.com>
Organization: Misanthropes-R-Us
Lines: 25

In article <9165@claris.com>, hearn@claris.com (Bob Hearn) writes:
> >C's grammar is not Context Free.
> grammar."  If you follow the technical definition, then there is no
> question that C's grammar (as specified in K&R) is context-free.

I think the crux of the argument of the non-context-free crowd is that
the grammar in K&R does NOT actually specify the C language by itself, not
as a grammar.  It relies on the trick of telling the lexer about typedefs,
which is a form of back-door context.

Consider:  if you wrote a YACC specification of C according to the context-free
grammar, but omitted any kind of action clauses (perhaps you want to turn on
debugging and watch the state machine play, a fine entertainment for a rainy
day), it would NOT be able to correctly parse a C program with typedefs (unless,
of course, you hid a "real" C grammar (with actions and the ever present
back-door gimmick) in the lexer itself).

When the pro-context-free crowd's argument is presented as "So what; it is
close enough, and it works", the argument in convincing.  When it is presented
as "Yeah, well, your mother wears army boots!", it isn't.
-- 
John Woods, Charles River Data Systems, Framingham MA, (508) 626-1101
...!decvax!frog!john, john@frog.UUCP, ...!mit-eddie!jfw, jfw@eddie.mit.edu

			Remainder Khomeini!