Message-Id: <199603240125.UAA12253@wilma.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: djb@koobera.math.uic.edu (D. J. Bernstein)
cc: drums@cs.utk.edu, moore@cs.utk.edu
Subject: Re: address syntax 
In-reply-to: Your message of "24 Mar 1996 01:08:09 GMT."
             <19960324010809.10349.qmail@koobera.math.uic.edu> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Sat, 23 Mar 1996 20:25:29 -0500
Sender: moore@cs.utk.edu

> > I'd far rather have an unambiguous grammar requiring at most one 
> > symbol of look-ahead when parsing left-to-right.  This would 
> > allow the ABNF to be trivially translated into yacc to construct
> > an automatic verifier that is guaranteed to be correct.
> 
> My ABNF can be trivially translated into yacc, right-to-left, to
> construct an automatic verifier that is guaranteed to be correct.

My ABNF can be fed to yacc in either direction.  

Most implementors are going to think in terms of left-to-right.  Should we 
make it difficult for them to do so just so they'll see how wonderful 
right-to-left is?

> > Neither should we simplify the syntax definition to require that parsers go 
> > from right to left.  Not everyone is willing to read and entire header field 
> > into a buffer in order to parse it.
> 
> I'm not sure who you're referring to, nor what the relevance is,
> since in fact they only have to read up to the next comma token.

Point taken.

Keith