I think that there should be normative language to the IESG to verify
consistency
In principle, I agree with you, but there are practical considerations.
First, I'm pretty sure we don't have IETF consensus to say that
working groups MUST carefully consider the registration policies,
coordinate them among similar/related registries, and write
justifications for their decisions... and I think we'd have a hard
time getting consensus for that. I think the document does the best
we can here, which is to explain why it's important, and to say that
this (English-word)should be done.
Second, few ADs have the interest in being rigorous about this, when
they consider the time spent looking at it, looking at other
registries, and arguing the point, and considering the other things
they're spending their time on (such as making sure the protocols
themselves are solid, that the security aspects are right, and so on).
When I was on the IESG, I did make a point of challenging working
groups on this point: please show me where the policy was discussed on
the mailing list, and please explain why you settled on Standards
Action, and why IETF Review or even Specification Required doesn't
work as well or better. Sometimes that resulted in changes, and
sometimes it resulted in reasonable explanations and no change. But
it took time to do that, and I think we can't expect it to happen in
general.
Third, while there is, as you point out, actual damage done by overly
strict policies (and I do point this out in the new text in the
document), the damage is limited and infrequent for the most part. A
few times, we've run into major issues with very active registries
(media types and port numbers come to mind), and we've put out
extensive documents that corrected those situations. But trying to
put MUSTs (whether they be capitalized or not) into a BCP to avoid
those is probably too much, and is not likely to get consensus.
Barry
-----Original Message-----
From: barryleiba(_at_)gmail(_dot_)com
[mailto:barryleiba(_at_)gmail(_dot_)com] On Behalf Of
Barry Leiba
Sent: Tuesday, May 24, 2016 3:49 PM
To: Roni Even
Cc: General Area Review Team; draft-leiba-cotton-iana-
5226bis(_dot_)all(_at_)tools(_dot_)ietf(_dot_)org; IETF discussion list
Subject: Re: Gen-ART LC review of draft-leiba-cotton-iana-5226bis
Hi, Roni, and thanks for the review.
I am wondering about the lack of normative language for example in
section
4.11
“When reviewing a document that asks IANA to create a new registry
or change a registration policy to any policy more stringent than
Expert Review or Specification Required, the IESG should ask for
justification to ensure that more relaxed policies have been
considered and that the strict policy is the right one.”
Is the “should” normative here?
Perhaps you're confusing "normative language" with "2119 key words":
text doesn't need 2119 key words for it to be normative, and this document
quite intentionally does not cite RFC 2119.
The example you give is a perfect one to show why we're not trying to
shoehorn key words that were meant to give instructions for interoperable
protocols into a document that's giving advice for writing and interpreting
IANA Considerations. The sentence above means exactly what it says in
English: it's advising the IESG, but it is ultimately the IESG's decision.
Barry