On 3/7/2008 10:56 AM, Russ Housley wrote:
So, I'll tell everyone how I deal with Gen-ART Reviews. Other
General ADs may have done things slightly different.
When I use a Gen-ART Review as the basis of a DISCUSS, I put it in
one of two categories.
(1) The Gen-ART Review was ignored. Like any other Last Call
comment, it deserves an answer. So, this is a procedural objection.
In this situation, I've been careful to say that the authors do not
need to accept all of the comments, but then need to answer them.
I have reviewed documents as a Gen-ART reviewer (during Brian's tenure
I think), sec-dir reviewer and also provided IETF LC comments on some
documents. As a reviewer, I am not sure whether I was expecting
answers all those times. I am pretty sure I have not always stated
whether or not the answers are satisfactory.
Next, I can imagine an author not wanting to respond to something I
may have said because it was totally bogus or inappropriate and does
not deserve a response. That might very well happen when I review
documents on a topic that I am not familiar with and haven't had the
time to read related references (that varies depending on the time
available, etc.). Perhaps that is not such a bad thing; being
blissfully ignorant on some topics keeps me, well, blissful. I use
somewhat of a hyperbole for obvious reasons. I am sure many other
situations are much more nuanced. I hope ADs don't continue to hold a
DISCUSS in those situations waiting for a dialog to take place or
waiting for a consensus to emerge. I sometimes hint in my reviews
that the topic may be at the border of my knowledge and if I have a
bias. Perhaps that is helpful.
Even if the response does not go to the person making the comments, ADs
need to see a response. Silence does not help us understand if
consensus has been achieved. Last Call is the only point in the
development and review of many documents where review from other IETF
Areas takes place. It is very important that this cross-Area review
Thank you for your note.
It's a fair thing to say that the ADs need to see a response. I also
agree that cross-area review is important and at times unearths issues
that may not have been raised in WG-level reviews. Personally, I prefer
cross-area reviews to take place prior to the LC process and hope that
the the LC process is for those issues that may have been really
overlooked despite the best efforts of the WG chairs and ADs.
I do not however quite understand the idea that we have to get consensus
in the context of each GenART/Sec-dir/<fill-in-the-blank> review. It is
of course quite plausible that one or two of those reviewers will never
be satisfied with any level of revision of a given specification. In
other cases, it may be that the reviewer has his or her personal
preference on how to write documents and will never come out and say
that the document they reviewed is "ready."
I have winced when some authors wrote to me after a review saying
something along the lines of "does the following revised text sound
better." I would have been merely pointing out something I thought was
not clear or might cause interoperability issues that they may have
overlooked or missed. How they might fix it is entirely up to them (or
the ADs involved). If their take was that they considered the comment
and thought what they wrote is much more meaningful in the context of
their work or the interoperability issue I raised does not apply within
the scope of their specification, well so be it. Next, I trust them to
do that. I don't need to see a dialog. I am willing to clarify my
comments, and if I cannot articulate the issues I am raising clearly,
well then the document needs to move on in the process.
Either the shepherding AD or the AD who solicited the review needs to
determine what if anything specific needs to be done in all these cases.
They are the judges of consensus. No matter who the reviewer might be
they were not selected by the community to do the AD's job.
Creating an environment where all these additional reviews and reviewers
essentially block document progress is not the right direction.
IETF mailing list