Date: 5 Dec 1996 06:06:41 -0000
Message-ID: <19961205060641.23404.qmail@koobera.math.uic.edu>
From: "D. J. Bernstein" <djb@koobera.math.uic.edu>
To: drums@cs.utk.edu
Subject: Re: Form of the Message Format document

> the standard should not be implementation instructions;

``Do this. Do that.'' What's wrong with ``Do what this code does''?

Look at the as-if principle in the C standard. Tremendously useful.

> but people can implement
> parsers in all sorts of ways; what you need is a good grammar to let you
> know if you've done it correctly.

``Need''? What for?

A message creator could feed sample outputs through a syntax checker,
which would test the MUSTs and SHOULDs. Is this the application you're
talking about?

Why is ``a good grammar'' the only way to provide such a checker? Are
you sure that religious adherence to a single tool is going to produce
the best result? Don't you think strcasecmp() might help?

Yes or no: Is DRUMS going to provide a usable syntax checker? If yes,
let's actually _do_ it. Right now we're not even close.

> As written, the grammar is more useful for writing a verifier than a parser.

I don't believe you. Show me a verifier.

> The problem with the way 822 is currently written is that there is no
> definitive way to tell where things like whitespace and comments are
> allowed.

RFC 822 states clearly that whitespace and comments are allowed between
tokens in structured fields.

I don't like the RFC 822 exposition of the tokenizer. ABNF really hurts.
Pseudocode would have been much easier to read and much easier to apply.
However, the spec is definitive; there's no dispute about how the
tokenizer is supposed to work.

In the high-level syntax, RFC 822 fails to explicitly state its notation
for _tokens_. The string "Mar" refers to an atom with contents Mar; the
string ":" refers to a colon token; the string 4DIGIT refers to an atom
whose contents are 4 digits. Fix: Define the notation.

> I go to 822 to answer the question "Can free insertion of linear whitespace
> go *here*?", where "here" is any number of places in a header, it's very
> difficult to tell.

A document written my way would make it crystal clear---and would only
have to say it _once_.

> Having everything in the grammar
> leaves no ambiguity, and having them in the prose is almost guaranteeing it.

Really? How come your grammar allows ``To: anything I want''? How come
your grammar allows the string ``foo'' to be parsed as three atoms?

---Dan
Put an end to unauthorized mail relaying. http://pobox.com/~djb/qmail.html