Re: [Gaim-devel] Jabber: user vs. user@domain

[Greg H. <ghudson@MIT.EDU>]

On Thu, 2005-11-17 at 08:48 -0600, Casey Harkins wrote:
> Since its not compliant anyways, would it be reasonable to require the 
> user to precede it with @, i.e. "@myserver.com/announce/online"?

myserver.com/announce/online *is* a valid JID.  It's
"myserver/announce/online" which is not (but which is apparently used
sometimes anyway).

It sounds like the practical applications of sending to a domain
generally involve a resource.  So I think we can safely and productively
translate a bare username (no dots, colons, at sign, or slash) into
username@defaultdomain without causing problems.  We wouldn't be munging
a valid JID and we wouldn't, to the best of my knowledge, be interfering
with any practical applications.

Objections?  I realize it would also be nice to handle usernames with
dots in them, but we can't do that without reinterpreting valid JIDs.
And while it might be convenient to send to user/resource without the
domain, that inteferes with sending to myserver/announce/online which
people apparently do.

(It's too bad the Jabber spec authors didn't anticipate this problem,
based on the net's experiences with email, and require the no-username
form of a JID to be "@domain[/resource]" like Casey suggests.  Then
there would be no ambiguity.)