Re: Thought on ABNF upgrade

* Julian Reschke wrote:
>So how do we proceed?

I think the grammar should not have any notion of implied LWS, I think
there are errors in the current grammar (I don't have the details handy,
sorry) and checking the prose to see whether there is implied LWS or not
is difficult and error-prone. Further, there should not be productions
where the rhs is described in prose, like <any US-ASCII digit "0".."9">.
I'd also like to see #rule rhs go, but that might have ill effects on
the readability of the grammar. Note that I'm just saying how it should
be written, not what the grammar should be.

>Optimally, we would then find a parser that will accept both 
>rfc2616-style ABNF and RFC4234-style ABNF, and which will re-serialize 
>into a single RFC4234-based format that we can then use to verify changes.

I wrote this some time ago, but I'm afraid the code is too fragile to
help much here. Note that such a tool needs human input as whether a
production has implied lws or not is hidden in the prose. Splitting the
grammar so that you have one set of productions with implied lws and
one without, along with removing <any US-ASCII digit "0".."9"> style
right-hand sides, would certainly help motivate writing a tool.
-- 
Björn Höhrmann · mailto:bjoern@hoehrmann.de · http://bjoern.hoehrmann.de
Weinh. Str. 22 · Telefon: +49(0)621/4309674 · http://www.bjoernsworld.de
68309 Mannheim · PGP Pub. KeyID: 0xA4357E78 · http://www.websitedev.de/

Received on Saturday, 17 March 2007 22:31:52 UTC