Re: [Tools-discuss] Nits in the idnits checker

[Henrik Levkowetz <henrik at levkowetz.com>]

Hi John,

On 2008-02-20 14:46 John C Klensin said the following:
...
>>> ### This actually is in the text, but it appears only in a
>>> ### comment in
>>> 	the ABNF. ###
>> Hmm.  Not sure what's right thing to do here -- but the ABNF
>> fix did
>> unfortunately not eliminate this one.  Should stuff in ABNF
>> comments
>> be ignored for this purpose, or not?
> 
> I'd be inclined to just let it go, i.e., continue to produce the
> message.  The use of references out of ABNF comments is very
> rare and probably not a wildly good idea.  The explanatory,
> rather than normative, BNF and ABNF of 821/2821/2821bis is
> unusual and less and less in favor these days and normative ABNF
> would, I think, use such comments even less often.  I've never
> had a discussion with the RFC Editor about it, but guess it is
> not something they would particularly encourage.   I've thought
> about heuristics that would make the test better, but,
> especially if we ever get the potential for numbered productions
> (for indexing purposes) in xml2rfc, I don't see any really good
> way to distinguish among
> 
>      ...
>                  ; see the production for FooBar
>                  ; see [RFC9999]
>                  ; see P15
> and even
>                  ; the optional construction 
>                  ;    [barbaz] was deprecated by [RFC9999]
> especially since, in a perfect world, one would want
>                  ; see RFC9999
> to generate an "unreferenced RFC" warning

Ok.  I'll think about this some more, but let it go for now.

>>>   -- Possible downref: Non-RFC (?) normative reference: ref.
>>>   '2'
>>>
>>>   ** Obsolete normative reference: RFC  821 (ref. '7')
>>>   (Obsoleted by RFC 2821) 
>>>
>>> ### But this is a reference to text found there that is still
>>> 	relevant, so the reference is not precisely to obsolete text
>>> 	### 
>> I looked at the text in the document.  Relevant I'll grant
>> you, but does
>> that mean it's normative?  For an informative reference, I'd
>> emit the
>> format you see below for 822, is maybe a bit more reasonable?
> 
> Yep.  
> In a perfect world --again, I'm not sure I'd recommend going to
> the trouble-- I would think the checker might want to remember
> any "Obsoletes" or "Updates" information in the document header
> and use that information to suppress "obsolete reference"
> warnings about those RFCs.  Informative references to previous
> versions are certainly more common than normative ones, but both
> occur.  However, if I identify the current document as a
> replacement for the older one in the header, I almost certainly
> know what I'm intending to do when I then reference it.

Right.  (Even if I'm not sure if an already obsoleted document needs
to be obsoleted again ;-)


Best,

	Henrik
_______________________________________________
Tools-discuss mailing list
Tools-discuss at ietf.org
http://www.ietf.org/mailman/listinfo/tools-discuss