Re: [abnf-discuss] constrained-01 - advantage?

[Sean Leonard <[e-mail]>]

Hello Doug:

Thank you...I think?

ABNF is already “machine-parsable”. A couple of months ago, I published (ss)reaex, which is a machine parser (extractor, specifically). Other people have written full parser-compilers (apg, bap). The fact that a compiler “can” be written for ABNF has always been there, although perhaps it was not a design goal of RFC x234.

I suppose that a goal of draft-seantek-constrained-abnf is to help authors make ABNF more machine-parsable, but, that is not the only goal. Another goal is to assist specification writers with making their ABNF clearer for (human) readers, which means introducing a minimum amount of new syntax for specification writers to learn. There are other IETFers who do not feel that machine parsing is a (strong) design goal. To that end, another design criteria is that ABNF still meets all the criteria of BNF: a context-free grammar notation.

There are other enhancements to RFC 5234 that have been proposed, which do take the language further:
RFC 7405
draft-seantek-unicode-in-abnf
draft-seantek-abnf-more-core-rules

Can you describe what you mean by “needing to go much further”?

Regards,

Sean

> On Nov 19, 2016, at 2:50 AM, Doug Royer <[e-mail]> wrote:
> 
> 
> I do not have a problem with making ABNF machine parsable.
> 
> This draft makes it possible to make ABNF that is still not machine parsable. Because the old difficult to parse methods are still valid.
> 
> In my opinion, if this draft is going to go forward, this draft will need to go much further, because its going to make a confusing mess of mixed ABNF usage in future drafts.
> 
> And much further, its going to be more controversial.
> 
> -- 
> 
> Doug Royer - (http://DougRoyer.US)
> [e-mail]
> 714-989-6135
> 
> _______________________________________________
> abnf-discuss mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/abnf-discuss