[jdev] mu-conference, JEP-0045 question/comments was "if anyone is having problems with slow propogation"

[Etan Reisner]

On Thu, 1 Dec 2005, Matthias Wimmer wrote:

> Hi Etan!
>
> Etan Reisner schrieb:
>
>> I don't want to hijack this thread, but since this one is active I thought 
>> I'd take advantage of the discussion of the mu-conference development 
>> status to request responses to my mail to standards-jig from Sept 29 about 
>> some JEP-0045 and mu-conference issues I have run into.
>
> You can fix mu-conference using this patch:
>
>>>>>> snap <<<<<
> diff -ur mu-conference-0.6.0/src/conference_room.c 
> mu-conference-0.6.0-fixed/src/conference_room.c
> --- mu-conference-0.6.0/src/conference_room.c   2003-11-15 05:19:45.000000000 
> +0100
> +++ mu-conference-0.6.0-fixed/src/conference_room.c     2005-12-01 
> 01:24:49.843414808 +0100
> @@ -273,7 +273,7 @@
>
>    if(room->secret != NULL)
>    {
> -       xmlnode_insert_cdata(xmlnode_insert_tag(invite, "password"), 
> room->secret, -1);
> +       xmlnode_insert_cdata(xmlnode_insert_tag(node, "password"), 
> room->secret, -1);
>    }
>
>    xmlnode_insert_node(result, node);
>>>>>> snap <<<<<
>
>> 
>> Please direct any responses to that email and not this one, and if I 
>> shouldn't have jumped in here please ignore me.
>
> Sorry, I do not have a copy of this e-mail anymore, so it's far more easy to 
> reply here than there ;-) But as it is just a bug in mu-conference, I do not 
> thing, that we need a big discussion on standards-jid about that.
>
>
> Tot kijk
>    Matthias

Unfortunately, that simple fix doesn't solve the problem. 1) The server 
I am working with is using ejabberd and as such uses the ejabberd muc 
component. 2) Fixing any single copy of muc isn't much help when all the 
other copies are still wrong, especially when as big and well known a 
server as jabber.org is using a broken version.

The real problem here is that the spec and the "reference" 
implementation disagree and I know that at least gaim had been tested 
against muc on jabber.org and as such was broken regarding the spec 
(which I noticed when it failed to work on our server at work). gaim 
in cvs currently obeys the spec but that causes a problem in the 
other diretion. The new problem being that client authors either need 
to pick one set of servers to be broken against (ones using 
mu-conference or ones obeying the spec), or they need to look in 
both places despite the spec existing and indicating the correct place.

Also there were a couple small other comments about the JEP in my email, 
but I'll send it again since I figure most people (liek you) don't have 
the old email anymore.

 	-Etan