scope of IDNAbis

[John C Klensin]

While I would prefer to avoid a discussion about what is
"allowed" and what is not, +1 to the rest of this comment.  More
generally, I think the tendency to try to use a particular
document or WG to "fix the Internet" just because work is
underway there and there is some relationship with the perceived
problem leads nowhere but into unneeded delays and rat hole
tours.

   john


--On Friday, 29 August, 2008 10:32 -0700 Erik van der Poel
<erikv at google.com> wrote:

> I don't think IDNAbis should have any MUSTs related to RFCs
> 3986 (URI) and 3987 (IRI).
> 
> IDNAbis should only use the key words "MUST", "MUST NOT",
> "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT",
> "RECOMMENDED",  "MAY", and "OPTIONAL" in sections that refer
> to the DNS protocol itself.
> 
> ftp://ftp.rfc-editor.org/in-notes/rfc2119.txt
> 
> For all other protocols, including URI, IRI, SMTP, HTTP, HTML,
> etc, IDNAbis should only give recommendations, without using
> any of the capitalized RFC 2119 key words above.
> 
> New versions of IRI, HTTP, HTML, etc should be the only specs
> that are allowed to capitalize MUST, etc with respect to IDNA.
> 
> Erik