[Standards-JIG] Broken mu-conference password invites?

Matthias Wimmer m at tthias.net
Thu Dec 1 10:29:22 UTC 2005


Hi Etan!

Etan Reisner schrieb:

> I'm resending this to this list because of some recent email on jdev. 
> For the record, I am aware of how to fix mu-conference to do the right 
> thing, that's not my issue. My issue is that the spec and the 
> "reference" implementation disagree, and I'm not best sure how this 
> needs to get fix.

Well I would not consider mu-conference to be a reference 
implementation. That might be the reason why I do not consider this a 
big problem. I'd just say: “mu-conference is broken, let's fix it”. As 
it really is a simple fix, it should be able to convince the admins (at 
least of the big servers) to install this patch.
As you pointed out in your private mail, there might be some clients 
looking for the <password/> element at the wrong place. You have two 
choices: do not care about that to make sure, that the users of these 
(broken) clients care to update their client, and that a fix gets being 
available; or just insert the <password/> element at both places, so 
that also clients looking at the wrong place will find the information 
they need.
For client development, it might be the right thing (if you do not want 
to break compatibility with buggy mu-conference-installations) to first 
look at the right place, and if there is no password, than have a second 
look at the wrong place inside the <invite/> element.

>
> Is there a reason that <password> and <decline> are children of the 
> <x> element but <continue> is a child of <invite>?


I think, they are included in the <invite/> element, as the invitor 
might not want to add this flag to the invitation to every JID he 
invites. … I don't think that changing that would make any sence, and 
just breaks existing software.

>
> And I don't really know how to read schemas but is the schema correct 
> for how the <continue> element is used/placed? It would seem it should 
> be under 'invite' not under 'item'.

Right …


Tot kijk
Matthias




More information about the Standards mailing list