Default Address Selection for Internet Protocol Version 6 (IPv6)
draft-ietf-6man-rfc3484bis-06

Thank you for a comprehensive Appendix B that made reviewing this
document much easier.

---

I note that the Ballot Write-up includes text from an old version of 
the Abstract saying:

   All IPv6 nodes, including both hosts and routers, must implement
   default address selection as defined in this specification.

This strong requirement appears to have been relaxed in the final
revision. Please update the ballot.

These comments are provided with the goal of improving the readability
and completeness of the information in the document.  I apologize for
the vagueness of my complaints about readability in comment 4; if
everyone else is able to read and understand the text I commented on,
feel free to ignore my suggestions.

1. I suggest adding a sentence explaining the status of IPv6
site-local unicast addresses and why they are included in the
procedures in this document.  Perhaps something like:

   "IPv6 site-local unicast addresses are deprecated [RFC 4291].
   However, some existing implementations and deployments may still
   use these addresses and, therefore, they are included in the
   procedures in this specification."

2. There is a minor conflict between the definitions of IPv6 multicast
scopes in RFC 4007 and RFC 4291.  I couldn't find an errata about the
issue.  RFC 4291 lists scope field value 0x03 as "undefined".  The
information kept by IANA seems to show RFC 4291 for multicast issues
in
www.iana.org/assignments/ipv6-multicast-addresses/ipv6-multicast-addresses.xml,
so, for consistency, this document might want to follow RFC 4291.

3. Minor editorial suggestion in section 4:

OLD:

   It is RECOMMENDED that the candidate source addresses be the set of
   unicast addresses assigned to the interface that will be used to send
   to the destination.  (The "outgoing" interface.)

NEW:

   It is RECOMMENDED that the candidate source addresses be the set of
   unicast addresses assigned to the interface that will be used to send
   to the destination (the "outgoing" interface).

4. The one part of this otherwise excellent document I find confusing
is the candidate source address selection procedure in section 4.  It
took several readings for me to understand how the rules in that
section are applied; e.g., which rules modify, override or are
considered in parallel with other rules.  I was unable to understand
this paragraph (which seems like it should be swapped with the
paragraph immediately following it for clarity):

   If an application or upper layer specifies a source address that is
   not in the candidate set for the destination, then the network layer
   MUST treat this as an error.  The specified source address may
   influence the candidate set, by affecting the choice of outgoing
   interface.

However, I'm not sure I have any good suggestions for broad
improvement.  Here are a few small suggestions:

   For all multicast and link-local unicast destination addresses...

   For site-local unicast addresses...

   Perhaps move the paragraph that begins "It is RECOMMENDED..." to
   the end of section 4 and begin it with "Unless otherwise specified
   above, it is RECOMMENDED..."

5. In section 5, the text describing the comparison rules specifies
three outcomes for each rule, "greater than," "less than" or "equal
to," while none of the rules explicitly defines an "equal to" outcome.
For completeness, I suggest adding a sentence explicitly identifying
the default result for each rule is "equal to".

6. Section 5, Rule 3: s/The addresses/If the addresses/

- Source address rule 5.5 has a lowercase "shall" - since it looks like
this document has had a fairly careful scrubbing of lowercase 2119
language I wondered if that's really a 2119 MUST or a "can"?

- Destination address rule 7 discussion: maybe s/6RD/6rd/ ?