On Saturday 11 November 2006 17:47, Steve Atkins wrote:
If nobody is ever going to use a piece of information, then expending
a lot of effort and complexity to make it available is a mistake.
Given that, plausible use cases are part of the process of deciding
what to keep and what not to.
Also, if there's a dependency on other standards for some functionality
(for example, a third party accreditation system of some sort) then that
dependency needs to be understood, so as to make sure that they
are flexible enough to work with some subset of reasonable systems.
(I don't see much along those lines with dkim-base, so none of these
issues should slow dkim development and deployment, but I do see
them all over with SSP).
This is all constrained by the quality of the crystal balls we all posses in
order to _know_ no one will ever use a piece of information. Given the IETF
track record with predictive use analysis, I don't think we should feel
particularly comfortable with the idea that we know what will and won't be
NOTE WELL: This list operates according to