On Wed, Mar 7, 2012 at 6:22 AM, Bill Fenner <fen...@fenron.net> wrote: > On Mar 6, 2012, at 5:44 PM, Brian E Carpenter > <brian.e.carpen...@gmail.com> wrote: > >> On 2012-03-07 11:26, Carsten Bormann wrote: >>> On Mar 6, 2012, at 23:08, Brian E Carpenter wrote: >>> >>>> Was there a real reason that you went for this? >>>> IPv6zone-id = 1*( unreserved / sub-delims / ":" ) >>> >>> I'm not Bill, but RFC 3986 says: >>> >>> IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" ) >>> >>> So you can't go outside that space without updating RFC3986. >> >> That inserts a very ugly v6. at the front, which is definitely not >> what people want. > > You're not going to give people what they want. What people want is > to use http://[v6addr%foo0]/, which I think we all agree is right out. > Once you've strayed from that, you're not giving people what they > want. > >> The point is practical: we want to stick a zone ID >> on the end of an IPv6 literal. > > Yup. We wanted to do that in 2004 too. > >> RFC 3986 is not sacred. > > Of course it's not sacred. That doesn't mean we should change it > without fully considering the ramifications. > > I think there are two issues here: > > 1. Should we use "%", "%25" or "something else"? These were the 3 > options in http://tools.ietf.org/html/draft-fenner-literal-zone-01, > and between the ipv6 working group and the URI working group we came > to rough consensus on the "something else" side. > What about "%%"? At least it's very close to the user's intention and models other ESC_ESC patterns (as in the C string "An escape sequence begins with a backslash \\").
> 2. Should we use the IPvFuture ABNF or modify the IPv6address ABNF? I > think there are much weaker feelings here - obviously it's easier to > do something that matches a Full Standard rather than modify it, and > for me personally the use case for this is so narrow (e.g., > configuring a home gateway for the first time via HTTP) that it is not > worth spending lots of cycles on it. On the concern of hosts using a > scope ID that does not fit in the IPvFuture grammar, RFC4007 says that > you SHOULD support a fully-numeric scope ID, so hopefully users will > be able to use the fully-numeric form. (Again, this goes to the rare > use case to me.) > It's perhaps not as rare as you believe going forward. We're supposed to be adding 50B (pick your favorite number) devices to the Internet in the next ten years (pick your favorite timeframe) and not all of them will have, nor be configurable through, a global IPv6 address. -K- > (Just a note about the use of "v6": the intent was to actually use > "v1"; the change log at the end of -02 claims that it changed from > "v6" to "v1" but the text didn't actually change. The version number > is meant to be the version of the IPvFuture, not the version of the > address inside it.) > > Bill > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > ipv6@ietf.org > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
