On Thu, Mar 11, 2010 at 10:31, Bernd Fondermann
<[email protected]> wrote:
> On Thu, Mar 11, 2010 at 09:51, Bernd Fondermann
> <[email protected]> wrote:
>> On Thu, Mar 11, 2010 at 09:06, Jean-Sebastien Delfino
>> <[email protected]> wrote:
>>> With ejabberd and gtalk (and IIRC jabberd too), if I send a message to
>>> [email protected] for example, and jdoe is listening through multiple client
>>> resources, like [email protected]/desktop, [email protected]/laptop,
>>> [email protected]/phone, he'll get a copy of the message on each resource.
>>>
>>> From what I can see in my test environment, Vysper seems to behave
>>> differently and only delivers the message to one of the connected resources.
>>>
>>> Is there a way to send a message to all connected resources like the other
>>> servers do? or did I miss anything in the way I've constructed the
>>> XMPPServer [1]?
>>
>> Looking at the code, this is the way it's implemented.
>> However, you could send a <message type="headline" > and this would go
>> to all resources.
>>
>>>
>>> I've checked the XMPP specs:
>>>
>>> XMPP Core - RFC 3920 Section 10.5 [2] says
>>> 'should deliver the stanza to at least one of the connected resources'
>>>
>>> XMPP IM - RFC 3921 Section 11.1 [3] says
>>> 'MAY choose between them or MAY deliver the message to all such resources'
>>>
>>> Then XEP-0259 [4] introduces a protocol extension allowing device resources
>>> to control which one should receive the messages.
>>>
>>> So, I'm not an XMPP expert... but unless Vysper implements XEP-0259 -- or
>>> another relevant XMPP extension from this spec maze :) -- I think it would
>>> make sense to send the message to all connected resources (unless they have
>>> indicated a negative priority, as described in [3].)
>>
>> ... unless of course you send to a resource directly.
>>
>>>
>>> Thoughts?
>>
>> I'm undecided. I'll go and figure.
>
> see 
> http://xmpp.org/internet-drafts/draft-ietf-xmpp-3921bis-05.html#rules-barejid-resource-message
> and VYSPER-179
>
> We really have the choice. Means we should give users the choice and
> make it configurable.
> After implementing the functionality, of course :-)
> Also, what is not working properly right now is how the prio
> resolution tie is resolved.
>
> I'll have a look.

In trunk, should now work as you suggested.

  Bernd

Reply via email to