Re: When to publish -12 - ABNF

Doug replied on 10/01/2003 06:08:43 PM:
> > For the base RFCs we had:
> >
> >    The memo also includes a formal grammar for the content type based 
on
> >   the Internet ABNF defined in [RFC 2234]. This ABNF is required for
> >   the implementation of parsers and to serve as the definitive
> >   reference when ambiguities or questions arise in interpreting the
> >   descriptive prose definition of the memo.
> >
> > but we dont seem to have anything like this in CAP. 
> 
> No one believes that there is nothing like ABNF in CAP.

Again I cant tell if you are purposely misreading what I said to draw 
things out or if you just didnt understand it.  I noted that CAP has NO 
paragraph in it like the cited one from 2445 that says the ABNF is 
definfitive over the text when interpreting the CAP text.  So when 
implementors try to implement CAP they have something to allow them to 
resolve issues.

Unfortunately CAP has nothing like this nor is the ABNF consistantly 
organized/formatted.  This means a CAP parser has nothing it can use to 
determine if the command / payload are properly constructed or not.  To me 
this is not goodness but perhaps Im the only one who thinks this...

> >  Why is that?  What good is provding ABNF that is not accurate?? 
> 
> It it accurate. If you want 100% of CAP to be in ABNF -  propose the 
ABNF
> to the list. Just like the query reply, some things just do not make 
> good ABNF.

The ABNF is not considered more definitive ("accurate") than the prose 
when issues or questions arise.  That was the main point.

In reponse to your "If you want 100% of CAP to be in ABNF" quip, we all 
should want this.  By that I mean that there is nothing described in prose 
that is NOT represented in ABNF somewhere in CAP.  Otherwise you could 
have text with no ABNF that can be used in implementations to confirm 
validity.  I dont see how that would be a good thing so perhaps you could 
explain how it would be.

> Yea right. So, I am supposed to do 100% of this work with NO proposals 
> at all.

If all you respond to is proposals or text that says "I propose" then Ill 
try that...

I propose that CAP Section 1 Introduction be changed to, somewhere, 
include the text:

   The memo also includes a formal grammar for the content type based on
   the Internet ABNF defined in [RFC 2234]. This ABNF is required for
   the implementation of parsers and to serve as the definitive
   reference when ambiguities or questions arise in interpreting the
   descriptive prose definition of the memo.

I further propose that the ABNF currently in CAP be checked for 
consistancy of layout.  The ABNF for commands seems to vary in style which 
is probably legacy from the multiple editors we've had.  It is not that 
burdensome a think to ask of the editor given how prolific he is with 
spinning out additional CAP addon drafts is it? 

Bruce
===========================================================================
Bruce Kahn                                INet: 
[email protected]
Messaging & Collaboration                 Phone: 978.399.6496
IBM Software Group                         FAX: and nothing but the FAX...
Standard disclaimers apply, even where prohibited by law...