Re: [nfsv4] rfc5666bis and rpcrdma-bidirection progress update

[David Noveck <[e-mail]>]

> There is a reason that 6 months is written into our standards process --
> that is the minimal time for complete delivery of an RFC changing the
> status of a draft, to Proposed and then to (now) Full Standard.  It's
> enough time for review.  We said so fairly clearly.

Certainly I've never seen a document published in less than this minimal
time.

Most documents take considerably more.  In some cases I'm not clear about
the reasons.

> There is a reason that we picked 24 months for failure to mature.

I'm not sure of that reason or what the implications of "failure to mature"
might be.

I've seen a lot of documents take longer than this.  The record is mixed:


   - RFC7530 took 71 months to go from -00 to publication.  It matured
   slowly but it did mature.
   - RFC5661 took 51 months to go from -00 to publication.  Given the size
   of the document, and the scope of the changes, I think that time is
   unexceptionable.
   - RFC5666 took 66 months to go from -00 to publication.  As we have
   seen, there are some problems with this document and it appears that it not
   get adequate review but I don't see that stopping things arbitrarily after
   24 months (in July 2006) would have made things better.  When people are
   rushing to meet a deadline, it tends to hurt the quality of review rather
   than helping it.
   - RFC7862 took 67 months to go from -00 to publication.  I feel that
   this length of time was the result of jamming together too many unrelated
   extensions, and the working group is now in the process of trying to
   address this issue going forward
   - draft-ietf-nfsv4-versioning will have taken 25 months to go from -00
   to the end of WGLC. It took a while to mature and I would have liked it to
   happen faster but I believe ithis document  should go forward promptly once
   WGLC is complete.

> When it comes to Last Call, Spencer is just a member of the WG.

Yes but we have often waited for him to start it.  Is that correct?  Should
authors, in consultation with rest of the working group, simply
schedule WGLC on their own.? Of course it may not be possible for others to
change the document state.

There are limits to the IETF's commitment to egalitarian ideals.  As Boss
Tweed said, "I don't care how the people vote, as long as I get to do the
nominating."

> His comments are due at the same time as everybody else.

True.  That implies that if he has issues with the document, he should send
them to the list.

On the other hand, RFC 4858 requires the shepherd to state, separately,
that he has read the document and does not know of such issues.  So it
appear that this sensible view is not shared universally.

> Shepherding is just prompting the IESG to get the items on their agenda.
It shouldn't take more than one person....

If you read RFC 4858 you get a different view.  Of course that document
says it is offered on an "AS-IS" basis, and might not be fit for any
particular purpose.

Some people assume shepherding is more extensive than you do.  Spencer
seemed to imply it was extensive but, once it became clear that this was
important, the needed documents were produced quite quickly.

My advice to authors is to to give a document a week waiting for a
write-up, and then proceed to extensive nag mail, as Spencer takes the lack
of such mail as indicating the write-up can be put off indefinitely.

On Wed, Nov 30, 2016 at 6:23 AM, William Allen Simpson <
[e-mail]> wrote:

> On 11/29/16 2:35 PM, David Noveck wrote:
>
>> Perhaps we can revisit your suggestions when some of the documents
>> currently in WGLC (three!) require shepherding.  For a very long time,
>> Spencer has been the only person doing shepherding and I don't feel
>> that this situation can go on forever.
>>
>> Speaking as the NSFnet oversight person who initiated the overthrow of
> the IAB because of their inability to approve RFCs and WG creation in a
> timely manner (6 months).
>
> And as a member of both the POISED and POISSON WGs that wrote the
> IETF process.
>
> There is a reason that 6 months is written into our standards process --
> that is the minimal time for complete delivery of an RFC changing the
> status of a draft, to Proposed and then to (now) Full Standard.  It's
> enough time for review.  We said so fairly clearly.
>
> There is a reason that we picked 24 months for failure to mature.
>
> When it comes to Last Call, Spencer is just a member of the WG.  His
> comments are due at the same time as everybody else.
>
> Shepharding is just prompting the IESG to get the items on their
> agenda.  It shouldn't take more than one person....
>
>