On 12/3/20 1:02 PM, Phillip Hallam-Baker wrote:
We vet new protocols for security, I see no reason to grandfather insecure legacy.
The reason, of course, is that people still have a need to use legacy protocols precisely because they're already deployed; and one person's legacy protocol is another person's essential protocol.
Also, we've extended many "legacy" protocols, some in better ways than others. If we've done so poorly with some particular protocol (FTP might be such a case), maybe it's worth revisiting to see whether the design choices made then actually turned out to work.
What Historic means in IETF terms is that IETF process has finished. There will be no further updates to the proposal.
Well, that's not how 2026 defines the term. But it strikes me as a useful distinction. Maybe we should understand that if a protocol continues to enjoy significant use (even if only in niches) that protocol should continue to be maintained. And if we decide to not maintain some protocol, labeling it as Historic might well be the right way to publicly declare that.
(another idea might be to delegate maintenance of some legacy
protocols to other parties - someone else is maintaining this now,
ask them if they still think it's useful.)
An applicability statement doesn't really have the same effect. But more importantly, how are you ever going to convince anyone IPv4 is dead if you won't even take finger to the woodshed?
I suspect IPv4 will continue to be used for decades after public IPv6 access is ubiquitous - on enterprise networks, on private networks, and to communicate with legacy devices.
But explicitly sunsetting the public IPv4 Internet in N years
seems worth considering. (Clearly operator groups need to have
some input into that.)
Keith
