[mailto:ietf-dkim-bounces(_at_)mipassoc(_dot_)org] On Behalf Of Jon Callas
Subject: Re: [ietf-dkim] Lean vs. Fat 'requirements'
On 10 Aug 2006, at 6:55 AM, Dave Crocker wrote:
Some of us believe, rather strongly, that this is a particularly
important "bias" to the development of the requirements list. It
occurs, to me, however, that it might not be clear whether there is
working group consensus on it.
I would be interested in seeing statements of preference for, or
against, having the requirements be minimalist, and include
only those
items for which there is clear rough consensus to include.
If an item engenders real wg controversy, it is *not* included.
Comments?
Minimalist good.
Blank checque vetoes bad.
The best way to make the requirements discussion interminable is to allow
groups to exercise a veto by filibustering requirements they dislike.
I don't care about the complexity of a requirements document. All I care about
is that the complexity of the solution be minimized.
I rate the skill of a protocol designer by their ability to find simple
solutions to complex requirements.
Time after time the IETF has produced worthless trash because the requirements
analysis was either incomplete or arbitrarily constrained. If you look at the
protocols that have become overly complex the root cause is frequently
inadequate requirements analysis in the first instance leading to protocols
that acrete a missmash of ad-hoc extensions.
The only reason that PKIX has CRLs, OCSP, SCVP, Attribute certificates all as
add on mechanisms to basic certs is because the original requirements analysis
was incomplete. The complexity of the resulting spec is enormous.
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html