[Standards] JID Escaping

Matthias Wimmer m at tthias.eu
Mon Jul 23 14:19:36 UTC 2007

Hi Peter!

Peter Saint-Andre schrieb:
>> JID escaping is when you want to have some "odd" thing in the node part
>> and you know what is node.
> Right. JID escaping is not escaping of full JIDs, it is escaping of node
> identifiers. A title "Node Identifier Escaping" doesn't sound as catchy,
> but we could change it if people really care. What we care about here
> are bare JIDs (which is what we assume people will exchange or talk
> about), but we ignore the domain identifier part since that is not of
> interest in this context (mostly, being able to re-use characters that
> people already use for email, like single-quote).

I think the discussion on JID escaping shows, that this XEP specifies
something, that still needs some fine-tuning - and that the current
definitions are not intuitive.

IMO JID escaping is good to have a standard way to map foreign
identifieres to JIDs, if needed for gatewaying identities. (Mapping
authentication identities from LDAP DNs, transports to other messaging
systems, ...)
But JID escaping (at least how it is defined at present) is bad for the
part where it tries to introduce new valid characters in the node part
just for allowing people to create accounts with funny new characters.

Personally I'd say that we should remove the complete part about JID
escaping in user interfaces. I'd prefere JID escaping only be done at
the above gateways and clients showing (only) the escaped addresses. JID
escaping should only be used where it is technically needed to address
or used identifiers, that cannot be changed because they are outside of

But for addresses we create inside Jabber (e.g. on Jabber account
registration) we should keep the characters limited to the characters
that are allowed inside the node part of a Jid. The characters allowed
there have been selected with reasons. If we start allowing to have
characters like the @ sign in what a user sees, and start that UIs need
to have separate input fields for node parts and domains, the usability
of our client software decreases.

Our design of addressing requirements should be driven by what a user
expects (and user's are used to just enter an address in the form of
user at server in a single field as they do it on their E-Mail clients, on
MSN, on SIP, ...) and by what results in a clear and intuitive UI. - And
not by the need of phreaks to have unusual characters in their addresses.


Matthias Wimmer      Fon +49-700 77 00 77 70
Züricher Str. 243    Fax +49-89 95 89 91 56
81476 München        http://ma.tthias.eu/

More information about the Standards mailing list