RE: CMC Comments?

["David P. Kemp" <dpkemp@xxxxxxxxxxxxxx>]

> From: "Phillip M Hallam-Baker" <pbaker@verisign.com>
> 
> That is ISO model standardization, not IETF model standardization.
> Remember the mantra:
> 
> 	'There will be NO flag days'
> 
> Backwards compatibility is what the INTER-net is all about. When you 
> have an installed base of 100 million odd clients you make sure that
> the standards proposed support them wherever possible.


The Web succeeded because it interoperates with the installed base.
Yet the http specification contains no conformance statement which
mandates that http implementations support ftp, gopher, or wais.

A minimally-conformant TLS implementation would not interoperate
with the installed base of 100 million odd SSL clients.  That is
an incentive for developers to support more than the minimum, not an
argument for raising the minimum.

Can you you provide an example of an IETF INTER-net protocol
specification which mandates support for a prior functionally-similar
protocol?  My mental model of the IETF says that interoperability is
achieved by designating a single mandatory-to-implement
protocol/algorithm/data-structure/whatever for each function, along
with a naming/versioning system which enables the negotiation
of alternative recommended, optional, or locally defined
mechanisms.