On Tue, Feb 21, 2012 at 02:56:41PM -0800, Murray S. Kucherawy wrote:
We like to see interoperability reports contain information about
features of a protocol that are used vs. unused, so that if and when
the protocol seeks advancement along the standards track, we can
decide whether we want to keep it in the revision.
I'd support Randy's point about implemented vs. used for an
"interoperability" report, but only because I think of the
"assessment" part as separate, not unnecessary.
Such assessments need to consider what's implemented, what's used, and
yes, probably some element of context for both. However:
Should we consider a protocol feature only used by abusive actors to
be one that deserves to be kept, or is only legitimate use worth
considering?
....is an overly simplistic way to pose the question.
Aside from the difficulty of judging "legitimate" use, there's the
question of why the feature is in the protocol in the first
place. Presumably protocol features exist because people anticipate
some net benefit from including them, and it seems that the right
question when deciding whether to keep a feature is not only whether
it's provided some benefit so far (i.e. it's implemented and used) but
also, if it hasn't, why not?
In answering the "why" question, "We haven't needed it yet but we
expect to in the following use case in the next 18 months" leads in a
very different direction to "so far, it's useful only to attackers
working against us" or to "the problem still exists but we have a
better way to solve it" or to "the use case never materialized in
practice."
A slightly broader discussion of how the real-world deployment of the
protocol resembles (or doesn't) the anticipated deployment seems more
useful than trying to answer the question you're posing here.
It's difficult to have such a discussion without ratholing, but it
seems to me that the same can be said of trying to define "legitimate"
use.
Suzanne
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf