[pim] /dino: IGMPv1 "retirement" and draft-eckert-pim-rfc1112bis discuss

[Toerless Eckert <[e-mail]>]

Let me start discussing the points brought up at IETF116 in email so it is
easier to track and also directly be able to propose text.

Dino was asking about what it should mean to remove IGMPv1 text from RFC1112bis,
and my answer on the mike @IETF116 (at least what i read in the minutes ;-) 
was maybe not enlightening as it should be.

So let me try to rephrase:

IMHO:

What i think we want is no further implementations of solely IGMPv1 by making it historic.
We did actually see solely IGMPv1 implementations even a few years ago in surveillance cameras
for exampl. 

All new IP Multicast host tsack implementations or updates should be IGMPv3 or RFC5790
 (and we can discuss IGMPv2 related guidance separately).

What i think we do not need it unnecessary churn: Aka: no need to remove
backward compatibility with IGMPv1 / IGMPv2 from IGMPv3 / RFC5790 implementations.
Aka: NO CHANGE WHATSOEVER in current and new implementation of IGMPv2/IGMPv3/RFC5790 because
of the change of status in IGMPv1 from full standard to historic. We can not throw
out that backeward compatibility before we have had a long enough time giving clear
evidence that IGMPv1 is obsolete. Which we have not done so far. 

So, maybe a statement like the following in RFC1112bis would help to clear this up:

----
Update to RFC1112

This document declares RFC1112 to historic. It superceeds all aspects of
RFC1112 except for the definition of IGMPv1 in RFC1112, Appendix I. This has
the following impacts.

All references to RFC1112 for the purpose of IGMPv1 do still need to
refer to RFC1112 and not this document. This specifically aplies to the mandatory backward
compatibility with IGMPv1 in IGMPv2 {{RFC2236}}, IGMPv3 {{RFC...}} and Lightweight IGMPv3
{{RFC5790}}: their behavior is unchanged by this document and even new implementations
need to implement it as required by those specifications.

All references to RFC1112 for the purpose of any aspects other than IGMPv1 (Appendix I.),
such as the overall way the host stack ("IP Multicast (ASM)") service operates and
how IPv4 multicast packets are mapped onto Ethernet are updated by this document
to refer to the definitions in this document.
----

Hope this works - process wise.

Any feedback welcome!

Cheers
    Toerless