Re: ICQ

[John F <johnf@mars.nettrek.net.au>]

On Wed, 30 Jul 1997, Travis Cole wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> 
> 
> On 31-Jul-97 Christopher Jason Morrone wrote:
> >I'd like an app that will announce to a central server that I'm online (or
> >maybe distributed servers that communicate with each other for better
> >stability), and that my client can check to see if my friends are logged
> >in, where they're logged in, etc.
> >
<<cut>>
> >
> >In any event, there should be an RFC, and the source should be free.  And
> >if it isn't written soon ICQ will take over the world. :)
> >
> >
> 
> This is also exactly what I would like.  Just a way to find out when and where
> my freinds are logged in so I can send them a message they will see imediatly,
> and the ability to chat and possibly send files to them.  This would also have
> to be a multiplatform system.
> 
> I would love to have this as in a free and open form but now the only option I
> can see is ICQ or AOL's AIM and right now AIM is the only one working for more
> than Windows.
> 
> If I were a programer (which I am trying to learn, but not even close yet) I
> would try to get something like this started.  But alas I do not yet poses the
> technical skills necessary.  Possibly there is someone who does and is
> interested.
> 
> I really think something like this would happen and I have a lot of ideas about
> it.  Please let me know if you wan't to talk more.

Could such a thing be based on how RIP works? ie participating
providers "broadcast" every 5 minutes idents of participating clients,
and this propagates to other participating servers.

I'm sure there would be a way of coding the information so that it 

1) has some sort of security
2) is conservative of bandwidth usage
3) protects the privacy of those who don't want to be involved

Such a scheme could also solve a major problem many face when it comes
to trying to track down the source of a security scare. This would
obviously have to be very tightly controlled.

Consider a set of servers that keep track of master databases of
participating users. Each email address would map to a code like

byte.byte.byte.255.byte.byte.byte.255

The first set of bytes would identify a network or a host, so it would
be from 1-4 bytes long. The second set would uniquely identify the
user (and would be of varying length according to the size of the
provider).

Now suppose jim@some.domain wants to know if bob@there.com is logged
on. jim makes a request of his provider, which asks one of the masters
for the code relating to bob. the masters flag that when bob is online
it will tell the server at some.domain. When bob loggs in, the server
at there.com tells the masters that 123.234.23.255.123.234.12.6.255 is
online. When bob loggs off there.com tells the masters that
123.234.23.255.123.234.12.6.255 is offline. 

The server at some.domain knows that if the masters report that
123.234.23.255.123.234.12.6.255 it will tell jim. If jim loggs off the
masters will be told by the server at there.com that jim
(234.123.45.67.255.1.2.3.4.5.255) is offline, and it will cancel any
pending notification to jim that bob is online.

Multiple notifications could be pooled into a single IP packet, so 1
packet could contain say 70-80 messages.

The4 masters could update to each other the maps from user names to
codes every 30 minutes. Each providers server would be responsible for
their clients codes (shades of DNS). the providers could run message
packets every 5 minutes.

There would be all sorts of issues relating to who pays for what, but
if the end users are paying $1/month for the service, then it should
cover the data costs at the very least.

Your thoughts?

John Foster


--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-user-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .