[Standards] Re: IDNA text for rfc3920bis

[Bruce Campbell]

On Mon, 16 Apr 2007, Matthias Wimmer wrote:

> Peter Saint-Andre schrieb:
>> "... this character MUST be stripped from the domain identifier before the 
>> JID of which it is a part is used for the purpose of routing an XML stanza, 
>> comparing against another JID, or constructing an XMPP URI."
>> 
>>> It should be made clear that this should happen at the same time as 
>>> stringprep, and is just another canonicalization step.
>> 
>> Agreed.
>
> In which case we just switched from not allowing the dot at the end to 
> considering both (with and without the dot) to be equivalent.

Not quite; transforming one to be equivilant to the other.

> As I said I am find with both ... but I thought we all tended to just not 
> allowing the dot at the end of a domain.

What should a client (following a service discovery tree) do when it 
receives:

 	<iq from="example.com" type="result" to="me at example.com/client" id="discoitems1" >
 	 <query xmlns="http://jabber.org/protocol/disco#items">
 	  <item name="Jabber IM server" jid="staff.example.com" />
 	  <item name="Jabber IM server" jid="users.example.com." />
 	 </query>
 	</iq>

Dave's MUST NOT paragraph says that 'users.example.com.' should not have 
been generated; it does not say what should be done when it is 
encountered.  (in this example, the XMPP routing fabric has no reason to 
reject or modify the stanza, leaving the client holding the bag as it 
were).

My personal preference is to silently ignore the sin committed by the 
example.com server, and proceed by stripping the trailing dot, with the 
client following the tree to 'users.example.com' .

-- 
   Bruce Campbell.