On 12/22/2015 4:51 PM, Eric Burger wrote:
We’ve got the existence proof of the counter example in SIP. I cannot count how many times I heard, “Yes, I know foo is broken, but it has been in 2543bis-02 and now we are at 2543bis-06 and we could not possibly make a change because so many vendors implemented to bis-02.” Egad! We could not change what was in an Internet*Draft*??? Imagine the pushback to changing an RFC
Eric,
Your anecdote demonstrates that some industry folk make choices entirely
independently of how the IETF labels things. Frankly, that
substantiates my point. The fact of such choices has always been with
us. We, and the world, cope with it.
The IETF has an obligation to provide clear documentation about the
meaning of our labels and to assign the labels according to those
criteria. That means the process of producing the document and the
process of approving publication.
We do /not/ have an obligation to anticipate people's using our
documents in ways that go beyond that labeling. And the fact that some
people make such choices puts us under no obligation.
It is a fact of life that specifications get changed or thrown away. A
company that makes an independent -- as opposed to community-based --
decision to make itself dependent on something that fragile so that it
can't cope with this sort of change has very basic problems and it is
not the IETF's responsibility to protect them.
The IETF's fundamental decision mechanism is community rough consensus.
No individual or organization has veto or approval authority. If the
/community/ wants us to retain something in an I-D or RFC, then we must.
If only an individual or organization does, we don't.
And none of this dictates the details of our quality assurance process.
What /should/ dictate that process is cost/benefit tradeoffs, not an
misguided, absolutist effort at perfection (which fails in any event.)
AD review appears to miss far more than it catches. That's seems to be
difficult for some folk to accept, and so they point to the individual
exceptions rather than the longer pattern of misses. (By way of some
very strategic examples, consider IPv6 and DNSSec and email security and
key management and...)
And AD review is extremely expensive. ADs are second-level managers and
reviewing is the task of an individual contributor. Having ADs do
late-stage reviews and having ADs spend significant time discussing
their views and preferences about the quality of this or that
specification is a strategic waste of valuable resources that should be
spent doing second-level management oversight.
In direct terms it adds time and frustration to IETF processes, produces
only random benefit while usually missing major problems, and reduces
the population of AD candidates due to the time-bloat of the job.
d/
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net