[Standards] Proposed XMPP Extension: Metacontacts

Kevin Smith kevin at kismith.co.uk
Tue Mar 27 12:56:57 UTC 2007


Pedro Melo wrote:
>> Well, what we decided was that it made sense to name the meta-contact 
>> the same way you would name the highest-order member
> 
> But the highest order member is not know unless all accounts are online.

The highest order member that's available to the client is known though, 
which is all that's required.

> Given that you want to make the protocol resilient, shouldn't we store 
> the name and groups in all account jabber:iq:private? 

Well, that's how we would end up with conflicts, which is why we've 
constructed the spec such that we don't end up with duplicated data 
which can then conflict.


> I mean, if you use the highest order member and you only have a partial 
> view of your accounts, you might end up with a contact that changes 
> group and name every time one of your accounts goes online and offline. 

That's true, but it's not a problem; it's a user choice how they name 
and group these contacts - if they want the different members of a 
metacontact to have different names, and different groups, then surely 
it's right to honour this?

> Also: how do you plan to take in account the <show> element? Do you keep 
> the order per 'show' value?

Which <show> element?

> We struggle with some of this with our client. We support meta-contacts 
> (our protocol uses only information available in the roster) for quite 
> some time now, but we are trying to improve the reaction to presence 
> changes.

Presence changes between the multiple accounts of a user, or of contacts?

/K




More information about the Standards mailing list