Harald Alvestrand wrote:
one nice thing about the schema/protocol being part of the naming scheme
is that it does *not* tie me to a single provider for all services - my
jabber service for harald(_at_)alvestrand(_dot_)no is provisioned from someone
who's got no relationship at all to my mail and web services.
There is an architectural 'trick' here, that I suspect is the key for making
thing homogenize in a way that is tractable:
The underlying specifications permit you to have different addresses, for
different services. They also permit you to have the *same* address.
So the fact that your jabber and email and... (whatever) services all get data
to you via "harald(_at_)alvestrand(_dot_)no" is an administrative choice, not one imposed
by some grand unifying architecture that needed to be designed perfectly from
the start.
The only "architectural" rule needed for this is to recommend that folks base
new adddressing on an existing scheme, to avoid collissions. For example, an
administrative rule that foo:harald(_at_)alvestrand(_dot_)no is only available for
registration to the recipient of mailto:harald(_at_)alvestrand(_dot_)no is all that is
needed to make this work.
(Anyone paying close attention will note that this introduces a problem with
getting a foo: address that is not the same as the email address but is not
assigned to anyone else. But what the heck, I'm not trying to design the whole
thing right now...)
At any rate, this is a version of the "think globally, act locally" approach to
architecture design that good Internet technical work did well.
d/
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf