Hi -
From: "Sam Hartman" <hartmans-ietf(_at_)mit(_dot_)edu>
...
Sent: Tuesday, September 05, 2006 8:52 PM
Subject: Re: Last Call: 'Procedures for protocol extensions and variations'to
BCP (draft-carpenter-protocol-extensions)
...
I want to be able to give you a URL and have you resolve it. That
only works if we speak the same transport protocol.
Sufficient, but not necessary. It wasn't *that* long ago that
transport-layer gateways were sometimes necessary to provide
the glue between IP-centric and other networking technologies.
I want people to be able to reference HTTP and get interoperability,
not to have to write a profile of http.
Defining the "MUST implement" features of HTTP results in a profile,
at least as the term was used in ISO / ITU-land.
In IETF-speak, describing the use of HTTP with a particular transport is
an "applicability statement". Including an applicability statement in a
protocol specification can be a reasonable thing to do. It's also reasonable
to talk about "interoperability" at the level of an applicability statement.
However, I think it would be a mistake to lose the current sense of
interoperability in terms of implementations' (lower) service interfaces.
Preserving those boundaries is essential to maintaining the modularity
of specification (and potentially of implementation) we need to keep
our sanity.
Randy
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf