Re: Comments for WD-css3-selectors-20010126

Bjoern Hoehrmann <derhoermi@gmx.net> wrote:

> >Rather than referring to RFC 1766, I would recommend to refer to
> >RFC 3066 [1] instead, which was just published recently and it
> >obsoletes RFC 1766.
> 
> RFC 3066 is a BCP whereas RFC 1766 is a Proposed Standard as per STD1.

And as I mentioned, RFC 1766 is already obsolete, as per RFC INDEX [3]
as of 2001-02-07.  STD1 (currently RFC 2700) needs to be updated.

> Is it really recommendable to make a BCP a (normative) reference here?

I do recommend it.  Otherwise, I'm sure the W3C Internationalization
Working Group will request it as a last call comment.

> I didn't see this before.

For example, HTML 4.01 Specification makes a normative reference to
RFC 2119 [4], which is a BCP.  CSS2 Specification makes a normative
reference to RFC 2318 [5], which is just an Informational RFC.

> Why is RFC 3066 only a BCP?

Indeed there was a discussion whether BCP is appropriate on the
ietf-languages mailing list.  Explanations from Harald Alvestrand,
an author of RFC 3066, were:

  From: Harald Alvestrand <Harald@Alvestrand.no>
  To: ietf-languages@Alvestrand.no
  Subject: Splitting RFC 1766
  Date: Fri, 04 Aug 2000 10:00:47 -0400
  Message-Id: <4.3.2.7.2.20000803101032.05f60e40@nordic.cisco.com>

  After consulting with the area directors, I have concluded that the Right 
  Thing for the next version of RFC 1766 is two documents, not one.

  One document describes the tag, its syntax and its registration procedures.
  This is the document that will be referred by all other specifications that 
  use it; it is standardized as a BCP.

  The other document carries forward the protocol elements of the current 
  draft, and will be called for Draft Standard (since it has seen heavy 
  usage, and is unchanged from previous versions).

  The reason for having the tag syntax in the BCP doc is to make the 
  dependency one-way; the protocol depends on the registration, and not the 
  other way around.
		<snip/>

and:

  From: Harald Alvestrand <Harald@Alvestrand.no>
  To: "Martin J. Duerst" <duerst@w3.org>, ietf-languages@Alvestrand.no
  Subject: Re: Splitting RFC 1766
  Date: Tue, 08 Aug 2000 10:55:56 +0200
  Message-Id: <4.3.2.7.2.20000808103938.00db1880@127.0.0.1>

		<snip/>
  The real reason I want the definition of the format to be BCP is because it
  makes life easier, because the BCP does not need to reference a document 
  that may have to cycle on the standards track.

  Formalistically, I can argue that interpretation of a tag has to be defined 
  once, and is by fiat, not experience; this is the same way that "assigned 
  numbers" is regularly reissued in revised versions without going through 
  the standards process. (If you gather that I think Assigned Numbers should 
  not be STD, you're completely right).
		<snip/>

You may look into archives of ietf-languages [6] for more details.

> [1] ftp://ftp.isi.edu/in-notes/rfc3066.txt
> [2] http://www.w3.org/TR/REC-xml#sec-lang-tag

[3] http://www.ietf.org/iesg/1rfc_index.txt
[4] http://www.w3.org/TR/html401/references.html#ref-RFC2119
[5] http://www.w3.org/TR/REC-CSS2/refs.html#ref-2318
[6] http://www.alvestrand.no/archives/ietf-languages/

Regards,
-- 
Masayasu Ishikawa / mimasa@w3.org
W3C - World Wide Web Consortium

Received on Friday, 9 February 2001 01:51:23 UTC