Re: The Evils of Informational RFC's

[Sam Hartman <[e-mail]>]

>>>>> "Bob" == Bob Braden <[e-mail]> writes:

    Bob> On 9/8/2010 3:12 PM, Richard Bennett wrote:
    >> It seems to me that one of the issues here is that architecture
    >> models are published as Informational when they're clearly not in
    >> the same level of authority as most Informational RFCs. An
    >> architecture document is meant to guide future work on standards
    >> track RFCs, and has been regarded historically as more or less
    >> binding.

    Bob> "...guide future work on standards track RFCs" -- yes.

    Bob> "...historically as more or less binding" -- no.
Bob, this was certainly an issue that came up when I was on the IESG.
At that time, we definitely felt that there were some architectural
decisions that the community as a whole had bought into.  We believed
that departing from such a decision was something that the community as
a whole needed to revisit.  For example, when a WG was chartered to work
on an architecture after the architecture document was approved, it
seemed fairly clear that the community had expressed a desire to have a
chance to look at that architecture.  Other times, however, it seemed to
us that a requirements document or architecture document represented the
thinking within a single working group. There, it didn't seem like
departing from this guidance required as much community review.

I'm summarizing a fair bit of discussions, but enough different
prospectives and examples were brought into the discussion that I feel
confident that while we don't know how large the sample size was, it was
more than just that IESG who believed there are times when architecture
documents are intended to bind.

I know I've often found the informational RFC label inadequate to
describe this sort of distinction and found that this distinction is
important to capture.

--Sam