[Standards] Re: IDNA text for rfc3920bis

Bruce Campbell b+jabber at bruce-2007.zerlargal.org
Mon Apr 16 13:26:40 CDT 2007


On Fri, 13 Apr 2007, Dave Cridland wrote:

> On Fri Apr 13 20:49:23 2007, Peter Saint-Andre wrote:
>> In my working copy I have added the following sentence:
>>
>>    If the domain identifier is a fully qualified domain
>>    name, the final character of the domain identifier MUST NOT be any
>>    character that is recognized as a dot, i.e., U+002E (full stop),
>>    U+3002 (ideographic full stop), U+FF0E (fullwidth full stop), or
>>    U+FF61 (halfwidth ideographic full stop); i.e., the domain name MUST
>>    NOT include an explicit zero-length root label as described in
>>    [STD13].
>> 
>> 
> I'm not sure redefining the list (even if it's the same values for now) is a 
> good idea.
>
> What about:
>
> If the domain identifier is a fully qualified domain name, the final 
> character of the domain identifier MUST NOT be a dot recognized as an IDN 
> label separator as defined in Section 3.1 requirement (1) of [RFC3490].

-1 for the above _as written_.

Firstly, what do you think will be interpreted by implementors as a 'fully 
qualified domain name'?  Number of (IDN) dots present?  Ability to be 
resolved?  Resolved where?

Secondly, what happens if the FQDN domain identifer does contain a 
trailing dot?  Do you reject the stanza at that point?  Strip the dot and 
forward it on?  Send it enclosed in another stanza to 'postel at isi.edu' 
(qualified by 'jabber:iq:ouija' of course) asking for advice?

What about domain identifiers that are not treated by implementors as 
being FQDNs?  Are they allowed to have a trailing dot?

My feeling is that the above is open to misinterpretation, and would 
suggest:

 	If the domain identifier includes a final character considered to
 	be a label seperator (dot) by IDNA (RFC3490) or DNS (STD13), this
 	character MUST be stripped from the domain identifier before being
 	used or forwarded further.  Ie, 'jabber.org.' becomes 'jabber.org'
 	and should be treated as being otherwise identical.

-- 
   Bruce Campbell.

"Be liberal in what you accept, and conservative in what you send."
 		-- RFC1122 paraphrasing Jon Postel.


More information about the Standards mailing list