[jdev] deferred delivery

Tijl Houtbeckers thoutbeckers at splendo.com
Tue Mar 1 07:37:39 CST 2005


On Tue, 01 Mar 2005 11:26:33 +0100, Mickael Remond  
<mickael.remond at erlang-fr.org> wrote:

> Tijl Houtbeckers wrote:
>> It's possible to use XMPP in a store and forward fashion, you wouldn't   
>> need to modify the protocol for it. It's still an "edge" case though.  
>> A  good reason for XMPP to replace email in my opinion is this concept  
>> of  "instant" error reporting.
>
> Yes. The idea of instant error is interesting, but the problem for what  
> to do with the error then rely on the client.
> If this is a simple chat message, then no problem, I wait for the next  
> time I see my correpondant online.
> But if this is an important message that I want him to read as soon as  
> he gets online ?
> I am using Jabber as some kind of text voicemail for people that are not  
> reachable.
>
> I agree however that I never encountered a Jabber server that was down  
> when I tried to leave a message waiting for someone.
>
> And I agree that the problem with mail is to process error message that  
> come afterwards. Clients has still to process the error message when it  
> comes as a definitive failure of delivery.

You can do two things.. modify an existing server to use  
store-and-forward, perhaps only under certain conditions (take a look at  
AMP, http://www.jabber.org/jeps/jep-0079.html ). Or you could write a  
component that delivers a message (if it's located on your own server, it  
could even spoof the from JID in most cases, take a look at the recent  
thread about that).

The latter idea would be more for when the user expclitly wants to use it  
to leave a message (I can imagine a client could try to deliver it the  
normal way, and when it gets an error offer to deliver it using that  
service). The advantage is it would (potentially) work on all jabber  
servers that support components, and people from other server could use it  
too (if you allow that, and without the spoofing ofcourse).




More information about the JDev mailing list