[JDEV] Jabber as COM/DCOM replacement for linux.

ogeorge at littledevil.com.au ogeorge at littledevil.com.au
Fri Mar 9 15:57:48 CST 2001


The more I think about it the more I love the idea.

It also highlights my least favourite part of jabber and offers a better 
solution.

I think interfacing say the chat transport through iq/set should not be an 
obscure flat naming thing (insert flame here) but instead should be an object 
oriented interface to a series of functions.  What i'm trying to say is that 
the interface should be XML-RPC across the jabber network.  

so notionally i might call functions on the chat transport by sending an iq/set 
(or perhaps an iq/xmlrpc) to a jid on the chat transport:
  <iq type=set to='chatt.localhost/groupid'>
    <query xmlns='chat'>
      <method>join</method>
      <args></args>
    </query>
  </iq>

look at www.xml-rpc.com/spec if you are interested.

Obviously this can be done already but it is the jabber way to standardise... 
has this train of though come up before?

Quoting Oliver George <oliver at littledevil.com.au>:

> 
> Jabber as COM/DCOM replacement for linux.
> --------------------------------------------
> 
> Think about it...
> 
> - you can access local/remote objects by a string identifier, the
>   transport JID
> 
> - you can define user/group privileges based on the sender JID (perhaps
>   mapped directly to unix user/group permissions)
> 
> - it has capacity for reference counting (presences)
> 
> - it has capacity for exceptions (responses of type="error")
> 
> - it has RPC functionality in the guise of IQ request/responses
> 
> - it allows language independent communication
> 
> - it would be really light/efficient where the packets are just passed
>   around by references to memory structures (and slower if it is
>   serialized in transit)
> 
> - Optional idea: it would even handle buffering messages where the
>   destination is not currently connected (bound to be useful for
> something?)
> 
> - Optional idea: you could use the XML-RPC type definitions for language
>   independent type passing.
> 
> I'd like to say at this point that a little knowledge is a dangerous
> thing and I'm definitely no expert. The idea seems really interesting
> though  :)
> 
> I'm curious as to anyone elses thoughts on the issue,
> 
> regards, Oliver.
> 
> 
> 
> _______________________________________________
> jdev mailing list
> jdev at jabber.org
> http://mailman.jabber.org/listinfo/jdev
> 




More information about the JDev mailing list