>>>>> On Mon, 19 Feb 2001 15:38:07 +0700,
>>>>> Robert Elz <[EMAIL PROTECTED]> said:
> | IMO, if we care about multi-sited nodes, the icmp-name-lookups spec
> | should be more scope-aware.
> Yes, if we want to handle that scenario. And it may be a good thing
> to add that extra bit of generality anyway.
> However, if, in your example, I1 had also had only site local addresses,
> then this scheme could not work in the first place. The assumption is
> that all nodes have to have global addresses. If we insist upon that,
> we may as well also insist that all interfaces on all nodes have global
> addresses (to be accurate here, this applies really only to nodes that
> are to be the targets of a DNS initiated conversation - nodes that only
> ever initiate connections, and never receive any, would not need global
> addresses. We should probably ignore that.)
> If I2 has a global address (G2), then the node info request using G2
> would find the site local (S2). And it may be that that is the desired
> result - if there's no site local corresponding to I1 (G) then someone
> who wants to communicate with G (ie: I1) probably should be using the
> global address (there had to be a reason I1 wasn't issued a site local addr
> after all).
Basically, I agree on all the points above, but I did not intend to
insist anything. I just would like to point out that it's not so
trivial to use icmp-name-lookups as a protocol of "tell me your site
local addresses". It is sometimes meaningless, and it sometimes does
not work...
JINMEI, Tatuya
Communication Platform Lab.
Corporate R&D Center, Toshiba Corp.
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
