[rfc-i] ABNF (RFC2234) vs HTTP's augmented BNF syntax (RFC822 + RFC2616)

Julian Reschke julian.reschke at gmx.de
Sat Feb 12 12:46:11 PST 2005


Bob Braden wrote:
> Julian,
> 
> I don't quite understand the question.  It would seem that if you
> invoke what you call the "implied *LWS" rule, then your "# rule"
> can be replaced by the "*" rule of RFC 2234 section 3.6.  What
> am I missing?

I think you're missing that the "#rule" (not mine, btw :-) also allows 
commas in the list production, thus

          1#element

translates to

          ( *LWS element *( *LWS "," *LWS element ))

which is clearly less readable.

> In general, standardizing on a single syntax notation seems like a
> very good idea.  Also, the RFC Editor checks every instance of
> ABNF using an ABNF checker, often revealing minor errors, BTW.
> If you introduce a new notation, will you give us a new checker
> for it? (And do we want to maintain two checkers? Not really...)

I completely agree with that goal. On the other hand, if the syntax that 
a spec describes frequently requires a shorthand notation like this 
(check RFC2616), what's the best way to resolve these two seemingly 
contradictory goals?

Best regards, Julian

-- 
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760


More information about the rfc-interest mailing list