Update to BCP-38?

Fri Oct 4 05:13:04 UTC 2019

On Oct 3, 2019, at 3:15 PM, Stephen Satchell <list at satchell.net> wrote:
> You still need a IPv6 version of RFC 1812.

If we were to start with the current draft, I would probably want to start over, and have people involved from multiple operators.

That said, let me give you some background on RFC 1812. The development started a little after a largish group of people started on what eventually became RFCs 1122 and 1123. The point was that there were a number of RFCs that needed to be updated with hard-earned wisdom - how ARP should work and so on. The group decided that instead of reissuing each individual RFC, they should do one omnibus effort. 1812 similarly covered a lot of "we discovered that this was true or needed to become true". The first editors was Philip Almquist, and it passed through several subsequent pairs of hands. It then sat on a shelf for several years, with people saying "we really should publish that" and not doing so. In 1994, the CIDR change was in full swing, and I was changing employers. Marshall Rose contacted me asking me to take it over, edit CIDR into it and "do whatever else was needed", and publish the thing. My new boss agreed to lt me do so, and I did.

I was given the text in RFC 1716 as input, which was then published as "historic", and RFC 1812 was the end product. RFC 1812 is kind of long in the tooth, BTW.

Since then, the way the IETF updates documents with hard-earned wisdom has changed. We don't do omnibus documents like RFC 1122/1123 and 1812; we write a document - or 20 of them - that "update" the document in question. You can find that information in the rfc-index.txt file, and in the datatracker. So if there were updates to include corresponding to those RFCs and on IPv6, I think the IETF would tell you to look at RFC 8200.

There is one thing in 1122/1123 and 1812 that is not in those kinds of documents that I miss; that is essentially "why". Going through 1122/1123 and 1812, you'll ind several sections that say "we require X", and follow that with a "discussion" section that says "we thought about X, Y, and Z, there were proponents of each, the arguments were X', Y', and Z', and we chose X for this reason". I would presume that what you're really looking for in a 1812-for-IPv6 is not "we require X" as much as "for this reason". Correct me if I'm wrong.

I can kick the idea around in the IETF if its important to you. I'll be looking for a LOT of operational input.