[JDEV] UIDs

[Scott Robinson]

Interleaved response.

Scott.

* Aaron Brady (insomnike) translated into ASCII [Wed, Sep 01, 1999 at 04:32:34PM +0000][<19990901163234.A336 at marvin.rage.org>]
> : 
> : Yes, with the current naming structure a local transport can provide
> : default names that are resolved to first in a case sensitive way.  This
> : allows you the flexibility in naming a large number of private transports,
> : without having to create DNS names for each one individually.
> : 
> : Of course, if your domain is foo.com, you can also just give them DNS
> : names and use @icq.foo.com and @robhome.foo.com.
> : 
> 
> If you look at IRC servers, most smaller servers are run by users, or fanatics
> within companies. The basic system this runs under is a) it doesn't use too
> many resources. b) it doesn't cause root hassle. Under normal circumstance,
> Jabber shouldn't use too many resources, but if you need to get your 
> hostmaster to add a new CNAME for every transport, you'll piss her off quite
> a bit.
> 

We've only decided that transports must be fully qualified domain names. Not
that they need to resolve to anything. =D In a previous post, we decided
that MX based routing would be how etherx resolved where to send the
message. A quick example:

scott at jabber.aspect.net sends message to 12341234 at icq.jabber.org

etherx at aspect.net parses icq.jabber.org and checks MX records.
Connects to jabber.org
Relays message to 12341234 at icq.jabber.org
jabber.org etherx relays to icq.jabber.org

The transport icq.jabber.org doesn't have to exist, it just needs the etherx
to have it in its routing table.

> If every transport had a small identifier, such as "icq", then
> icq://jabber.phishy.net/13141515 could be sent to the etherx on jabber.phishy
> and it could look up in it's local list of transports which (if any)
> correspond to 'icq'.
> 

I initially proposed this, but have dropped it after the issue that it would
conflict with the current URL RFC. We can only use "jabber://" with any
sanity.

[snap]
> 
> For this to work well, we should have a central list of widely used transport
> identifiers, like Creator ID's with Apple/Palm, so you don't some people with
> aim:// and some with aol://.
> 

If we _did_ do it this way, the transport identifier wouldn't have to be
unique.

> Also, for the more used tranports, a DNS record at jabber.org like ->
> 
> icq	CNAME	jabber.phishy.net
> 	CNAME	jabber.mirabilis.com
> 	CNAME	icqtrans.earthlink.net
> 	...
> 
> So by using the easy to remember icq.jabber.org, ICQ messages would be load
> balenced. Unfort, this makes getting replies BACK an issue.
> 

This can be referenced later. If there is no MX record, icq.jabber.org (in
our previous example) would have to resolve out to be routed.

[snap]