I'm not sure we want to throw out the idea of 'scope'. Even without SL, two scopes are architecturally enshrined in IPv6, link-local (addresses are valid and unique only on-link) and global (addresses are globally unique).
You'll notice that I don't say addresses are globally valid. Architectural scope talks about uniqueness. Functional scope talks about where an address is actually useable - the presence of filters and the such mean that there are many situations where an address is unique but wont work. This is a feature, not a flaw. The problem with SL and scope were twofold, and both were about non-uniqueness. A single SL address space is robust, even if nested within a larger global address space, as long as the address selection algorithms have a mechanism for evaluating scope and choosing an appropriate address. However, there is no guarantee of uniqueness between SL spaces, which creates problems on merge. Worse, SL allowed non-unique spaces to overlap and required an additional 'scope identifier' to determine which space an address belonged to. This scope identifier became a piece of addressing information that was no included in the address, with obvious problems. Given that we have a mechanism for solving the uniqueness problem (see note 1), then I think scope is a healthy concept. At least for address selection, its not that hard to encode certain scopes to certain address ranges and have address selection respond accordingly. Dealing with scope in a more general sense (eg I know this address - is it valid in any of my scopes?) is a knottier problem, but that problem exists regardless of whether scopes are defined. Having a mechanism to map an address to a scope makes the above problem easier. Note 1: Link local and the proposed local addressing draft both propose mechanism to cause the addresses generated to be unique, even outside their scope. LL uses the MAC address, while local addressing uses either allocation or randomization. -- Andrew White -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
