Ran, Going directly to your conclusion, it appears we have a clear common interest!
Each of the designs we are interested in depends, to be complete, on reservation of a subset of the IID space left unused by RFC4291 (that having u=g=1). Since you talk about "a small portion" of this space for ILNP, and since 4rd only needs only 1/2^14 of this space, I rather see convergence than conflict. Hoping we can agree on it, Regards, RD A few secondary comments inline. 2013-02-01 14:54, RJ Atkinson <[email protected]> : > > On 31 Jan 2013, at 13:11 , Rémi Després wrote: >> What ensures 4rd doesn't conflict with ILNP isn't at all >> that ILNP only uses u=0. >> >> It is that, in ILNP, no u=g=1 is used in *unicast* addresses >> (those whose IIDs are specified by RFC 4291). > > This is still inaccurate. > > I REALLY would be greatly obliged if you would not speculate > about ILNP on IETF/IRTF mailing lists, primarily because > any speculation is likely to be wrong -- and create needless > confusion amongst other folks on those lists. Contributing on IETF mailing lists, including on what concerns existing ILNP RFCs, is a right anyone has AFAIK. > Some published ILNP papers talk about a form of multicast > addressing/routing that we believe is novel and are exploring. > This combines a unicast routing prefix in the high-order > 64-bits with an IEEE EUI-64 compliant (and RFC-4291 > compliant) multicast group ID (i.e., U=G=1) in the low-order > 64-bits. > Because of existing IETF standards-track work > where U=0, for example CGAs or "privacy" addresses, the > U=0 identifier space (i.e., U=0 and G=1) can NOT be used > for multicast identifiers. This is precisely the point I have actively contributed to make in 6man! > So the proposed 4rd reservation of all (or most) uses of > addresses with the combination of (A) U=G=1 identifier space > and (B) unicast routing prefix is a direct conflict with > published ILNP papers and active ILNP work. The currently proposed IID prefix is 0x0300. It only uses 1/2^14 of the u=g=1 space of RFC-4291. This can't be characterized as " reservation of all (or most) uses" of this space. (Your turn to misrepresent, by mistake, the situation. Your apologies after mine will be welcome.) > Both ILNP and 4rd are Experimental, at least today. > This is why I tabled the possibility of allocating > a small portion of the U=G=1 space under RFC-3692 shared > experimental use rules. That would permit multiple > experiments to proceed, and is the usual IETF custom > for experimental work when a limited protocol-registry > resource is involved. As said at the top of this mail, this shows we have converging interests. -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
