On 10/05/2013 01:13, John C Klensin wrote:
>
> --On Thursday, May 09, 2013 03:32 -0500 Spencer Dawkins
> <spencer@xxxxxxxxxxxxxxxxx> wrote:
>
>> So in this case, we're looking at "RFC Editor state" =
>> "Heather, please do something" + "some working group, please
>> do something" + "author(s), please do something", and we can't
>> tell how much time to attribute to each of these ...
>
> You could further add to that list "RFC Production Center,
> please do something" (different from an RSE wait, which is, or
> at least ought to be, more significant) and "IESG or appropriate
> AD, please do something", which does happen.
>
> But the RFC Editor's numbers try (almost always successfully) to
> separate the two "waiting on the RFC Editor Function to do
> something" (Heather plus Production Center plus, in principle,
> Publisher) from the other states. Those other states could,
> from their point of view, be aggregated into "stuck, someone
> else's problem". If we are looking for issues with IETF
> end-game process, we need to parse those, but that is a
> different sort of question in terms of data-gathering and
> reporting.
All of which suggests that, ideally, the tracker would include
a variable "onTheHook" for each draft, containing at all times
the person or role responsible for the next step. That isn't
necessarily implied by the state machine. For example,
AUTH48 doesn't always imply that the author is on the hook -
it may be that the author has asked the WG Chair to ask the
WG for a quick review of a proposed last-minute change. At
that point, the WG as a whole is on the hook.
Brian