Lookup CONTEXTJ test

Simon Josefsson simon at josefsson.org
Sun Jan 9 21:52:48 CET 2011


John C Klensin <klensin at jck.com> writes:

> --On Sunday, January 09, 2011 16:22 +0100 Simon Josefsson
> <simon at josefsson.org> wrote:
>
>> Vint Cerf <vint at google.com> writes:
>> 
>>> simon:
>>> 
>>> from RFC5894
>> 
>> Thank you, that was the explanation I was looking for.
>> 
>> I note that RFC5894 is not a normative reference from
>> IDNA2008, nor a standards track document, so if we want the
>> explanation in that section to have bearing on
>> implementations, that should be fixed.
>
> Simon,
>
> As the RFC 5894 text that Vint quoted indicates, the actual
> rules are in RFC 5982, which is normative.  I've just looked at
> those rules again and I think they are adequate and clear (but I
> might be too immersed in that text for my judgment to be
> reliable).  If that is not the case, then something clearly
> needs to be fixed but the intention is that 5894 provides an
> overview and easy-to-read explanations while 5890 through 5893
> provides the specific and normative rules.
>
> As an implementer, you should probably be looking closely at
> 5890 through 5893 only (plus or minus whatever you want to do
> about mapping) and looking at 5894 only for help in navigating
> the documents or understanding the intent of the other ones. 

Agreed -- and this is exactly my point.  If we want text in RFC 5894 to
have bearing on implementations, it has to become normative.  There is
no definition of what a "null rules" is in 5890-5893.  That makes the
following hard requirement in 5891 problematic:

   To check this, each code point identified as
   CONTEXTJ or CONTEXTO in the Tables document [RFC5892] MUST have a
   non-null rule.

Based on this thread, I'm proceeding with my implementation treating the
explanation in 5894 of what a null rule is as the normative definition.

Thanks,
/Simon


More information about the Idna-update mailing list