[JDEV] The "OpenAIM" Project

Michael F Lin MFLIN at us.ibm.com
Tue Jan 8 23:15:58 CST 2002

I would just like to throw the idea of null clients into this mess. You
will recall that a sendmail null client is an instance of sendmail
configured such that it will not relay any external messages, but it will
send messages originating from the local host. This translates into Jabber
as a local daemon that looks like a Jabber server on localhost to any local
processes (clients, apps, aim-t), but connects out to the Jabber cloud as a
client. In the process of routing, it would rewrite the JIDs, just as the
sendmail null client rewrites the email addresses. Since it is not
externally a Jabber server, AOL can't log into it.

Now obviously this "violates the spirit of jabber" in that we need a much
heavier implementation (essentially a Jabber server) on the client machine.
However, this solution is compatible with existing client software, and I
would also like you to consider some of the possibilites it opens. Firstly,
it would allow Jabber to be used as a local IPC mechanism, which could be
quite interesting, and would scale very naturally to an RPC mechanism.
Secondly, (this idea is also copied from sendmail), it allows for
end-to-end store and forwarding, which IMHO is something we will need down
the line anyway in order to viably build web services on top of Jabber.

Obviously this idea is still a rough cut, but take it into consideration.


                      <jeremie at jabber.o        To:       jdev at jabber.org                                                               
                      rg>                      cc:                                                                                     
                      Sent by:                 Subject:  Re: [JDEV] The "OpenAIM" Project                                              
                      jdev-admin at jabber                                                                                                
                      01/08/2002 11:43                                                                                                 
                      Please respond to                                                                                                

There is a very simple reason why any approach like this is doomed to

As far as I can tell, the technique AOL is/was using to track the IPs
being used by aim.jabber.org was that they had a dummy AIM account and
registered the transport to use it via jabber like any normal user does.
Either they just used a normal client to do this or script with a jabber
module. Once aim.jabber.org logged into AOL w/ that dummy account, they
would examine the source IP for that client connection (I'm sure they
regularly block abusive users/IPs and this action is quite easy with their
administrative tools).

The entire thing could be automated on their side, and it would only take
a very short amount of time to obliterate any network of aim transports or
socket redirectors.

I fully agree, this battle isn't one for the technology, our technical
resources are better spent improving Jabber, and if anyone has political
resources those are probably best spent showing the world why a commons
for communication platforms is so important over closed
commercial/corporate networks.


On 8 Jan 2002, Adam Theo wrote:

> Hmm... A thought just occured to me when reading about these Socket
> Redirects. I am not familiar with them, so they may already have this
> ability.
> The key would for any "permanent solution" to be completely transport
> side. This is opposed to the client-side which would require users to
> install new software (won't happen), or even server-side which would
> require server admins to re-do their entire server installation. Here's
> a solution:
> Modify Temas's AIM-T to find other AIM-T's on the Jabber network in a
> DNS-like propogation system (how DNS entries spread accross the
> internet). When someone connects to an AIM-T, any AIM-T, the collective
> AIM-T's "shuffle" the users connections around, randomizing IPs and
> distributing load. Once a hundred or so IPs are on this "OpenAIM"
> network, it would be near impossible for AOL to track down even a small
> percentage of the IPs... especially if the IPs are somehow transparent
> to the client (to stop an AOL employee downloading and tracking AIM
> connections through Jabber). The only IP the client would see is the
> AIM-T at their home server, but the IP that actually is making the
> connection could be any one of dozens if not hundreds. Alot of potential
> here, folks... And this OpenAIM network would bring on alot of those
> "multi-protocol" clients that are not yet 100% Jabber... I would see
> Everybuddy and GAIM becoming full Jabber clients if we could pull this
> off...
> And in actuality, I think alot of the technology to do this is already
> out there, it just needs to be pulled together.
> Yes, I'm 100% behind this idea. I am a crappy programmer, but I would be
> willing to dedicate some pocket money to help a programmer or two get
> this up.... Whadda say? I know there are some problems, but instead of
> shooting this idea down, how about we put our thinking caps on and
> figure out viable solutions? Wow, I think this could work...
> _______________________________________________
> jdev mailing list
> jdev at jabber.org
> http://mailman.jabber.org/listinfo/jdev

jdev mailing list
jdev at jabber.org

More information about the JDev mailing list