Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!megatest!djones
From: djones@megatest.UUCP (Dave Jones)
Newsgroups: comp.lang.misc
Subject: Re: Language Design
Message-ID: <3015@goofy.megatest.UUCP>
Date: 23 Mar 89 02:53:28 GMT
References: <9140@claris.com>
Organization: Megatest Corporation, San Jose, Ca
Lines: 58

From article <9140@claris.com>, by hearn@claris.com (Bob Hearn):
> In article <2970@goofy.megatest.UUCP> djones@megatest.UUCP (Dave Jones) writes:

> 
> If you *had* been following the discussion, you'd have seen that my
> original posting was a followup to someone complaining about how
> screwed-up C's grammar was, that it wasn't context-free.  When someone
> uses the words "context-free" and "grammar" in the same sentence, I
> assume he knows what he's talking about, ...

What a silly assumption.  I almost never assume that people who post
to the net know what they are talking about.  But what you really
mean, I think, is that you want to use the term "context-free" exactly
the way Chomsky did.  Okay, that's your right.  But don't be surprised
or offended if someone doesn't know it.  Just explain.

I taught two terms of formal language theory during a two year stint 
as Visiting Associate Prof at Ohio University.  I'm sure I brought the
subject up, but only to try to avoid the kind of discussion that is
going on here. We never talked about (Chomsky) context-sensitive
grammars again.  I suspect that for most students, that small piece of
information got lost along the way.

Context-sensitive grammars, (in the Chomsky sense), are just not useful.
I really don't know why people continue to use his definitions.

I just now pulled a new compiler book off the shelf, completely at random.
It does pay lip-service to the Chomsky definitions. But  of its 811 pages,
it devotes less than one quarter of page 97 to context-sensitive
grammars!  First it gives a very brief definition, then the following:


      While context-sensitive grammars and type-0 grammars are more
      powerful than context-free grammars, they are also far less
      useful. The problem is that efficient parsers for these extended
      grammar classes do not exist, and without a parser there is no
      way to use a grammar definition to drive a compiler.

      [ Two sentences about parsers for context-free grammars omitted... ]

      Whenever we mention a grammar (without saying which kind), the
      grammar will be assumed to be context-free.


> [...]
> I've received no end of mail and flames for that simple correction!

Take my advice. Quit worrying about the flames you receive. Flamers
are small minded bigots, pedants, and idiots. Just ignore them.
(Or better yet, *apologize* sincerely.  God, that irritates them!)



                        Dave J.


P.S. Spring has sprung.  I probably won't be around the net much
in the next few weeks.