--On Monday, December 21, 2015 20:52 -0800 Dave Crocker
<dhc(_at_)dcrocker(_dot_)net> wrote:
On 12/21/2015 5:59 PM, John C Klensin wrote:
I'm not certain we are getting to "good enough" quite often
enough. Moreover, what was "good enough" when the expectation
was that people would not deploy Proposed Standards in
products, at least without understanding that was a risk and
treating it as such, may not be "good enough" when Proposed
Standards are not only deployed but the community's attitude
seems to be that
This nicely summarizes a common bit of mythology in the IETF.
First it presumes that folk out there in develop-and-deploy
land have no ability to assess what they are developing and
deploying.
Second is that it presumes that there has been some sort of
major change in the way IETF specs are processed pre- and
post- Proposed status assignment.
Both are fundamentally wrong.
The IETF is a collaborative community venture, not a grand
parental oversight commission. Folk out their in product-land
have been able to deal with immature, flakey and changing IETF
specs productively for more than 25 years.
Dave, I can (and do) agree with the comment in your last
paragraph without agreeing with the previous two (although we
could debate what changes are "major" or what "fundamentally"
means).
Perhaps I was just unclear because I wasn't talking about "the
way IETF specs are processed" but about how IETF specs at
Proposed Standard are interpreted in the community and how the
IETF responds to those interpretations.
I'm personally very sensitive to the change in the latter case
because, when I first took over as an Apps AD (you might recall
it was a bit mid-term -- information that is long enough ago to
be irrelevant to most others), the very first task I had to deal
with involved a major vendor who had deployed a product based on
a either a very late I-D or a proposed standard spec (don't
remember; could go look it up) after which the IETF changed its
mind, decided the original spec was a bad idea, and specified
something different and very incompatible with that original
idea. They felt that their having deployed the original spec
should require the IETF to stick with it, bad idea or not. I
got to explain the IETF's conclusion that the original was a bad
idea even though they had gotten it to work, tell them that the
situation was their tough luck for deploying at scale that
early, and that, unless they made changes, they simply were not
going to interoperate with anyone who did follow the spec.
For better or worse, I have a lot of trouble believing that
scenario could play out the same way today. Again, whether that
is a major change or not could be debated endlessly, but it is a
change.
best,
john