[nvo3] Comments on draft-hy-nvo3-gue-4-nvo-04

"Murray S. Kucherawy" <superuser@gmail.com> Tue, 28 March 2017 22:39 UTC

MIME-Version: 1.0
From: "Murray S. Kucherawy" <superuser@gmail.com>
Date: Tue, 28 Mar 2017 17:39:01 -0500
Message-ID: <CAL0qLwY36+yi-hWE7+jq22uJcs9ThRdYF80isSOhGoOHgunnjw@mail.gmail.com>
To: nvo3@ietf.org
Content-Type: multipart/alternative; boundary="001a114dbd7064d018054bd22086"
Archived-At: <https://mailarchive.ietf.org/arch/msg/nvo3/6t6ZIu-vlO_GdfT-uEgzmE1XZNg>
Subject: [nvo3] Comments on draft-hy-nvo3-gue-4-nvo-04
Precedence: list

I've been asked to review various NVO and GUE documents, and this is my
next target.  Hopefully these are helpful.

Section 1:

1) The [GUE] reference is stale.  The referenced document is now an intarea
work item.

2) "same was" should probably be "same way"

3) "applications, that" should be either "applications that" or
"applications, which"

Section 3:

4) "after UDP header" should be "after the UDP header".

5) Some of the field names aren't capitalized when they should be.

6) The last two paragraphs are repetitive and probably not needed.  They
all seem to say that the package MUST comply with GUE and NVO3.  I think
it's enough to say that once, but I also think you don't need to say it at
all given that it's clear from the document already that these are done in
the GUE and NVO3 context.

7) There are instances of "may" scattered around in this section and below
that need to become MAY or some other word (RFC2119).

Section 4:

8) Is that first "MUST be" necessary?  Could you just say "are"?

9) The next two MUSTs are possibly not needed, for the same reason as in
(6) above.

10) The subsections of Section 4 are indented.  Was this prepared with some
non-standard mechanism?

Section 4.1:

11) Could "MUST set" just be "sets"?  Could "MUST use" just be "uses"?

Section 4.3:

12) Given that SHOULD (see its definition in RFC2119), what's a situation
where one might legitimately not do what the SHOULD says?  If there is no
such case, make it a MUST; if there is such a case, please describe it.  If
you make it a "MUST", you can't use "avoid", because "avoid" leaves wiggle
room.

Section 5:

13) I think this has to be done more clearly.  There are two actions to be
taken here, and they should be described individually, preferably in their
own subsections, and referencing by name the two registries that are being
accessed.  You might include a normative reference to the document where
the registries are created, which I think is [GUE].  See RFC5226.

-MSK

[nvo3] Comments on draft-hy-nvo3-gue-4-nvo-04 Murray S. Kucherawy