Hi Margaret. > Is it optional for a vendor to implement temporary addresses?
That is my understanding, and I'm sure the topic will come up in the context of the general node requirements document. Note that the cellular requirements document we have been working on has wording now that says "should". > Is it optional for a user to configure site-local addresses on a box I'd say yes. There certainly can be no requirement to use them... > (or perhaps even for a vendor to support them)? What does it mean to not implement site-local addresses? Would an implementation reject any packet with a site-local prefix in it? For a host, I think you'd be hard pressed to not support them. For a router, I'd assume one has to implement configuration stuff to enforce boundaries. > One reason that has been put forth for having a fixed set of address > selection rules is that it will be possible for implementations to > know what addresses will be preferred, and override that behaviour > if desired. > I'm not quite sure how this works when it isn't clear that all > vendors will implement all address types (temporary, site-local, > etc.), or that all address types will be configured for a given > system. This particularly becomes problematic when optional address > types (temporary, perhaps site-local) are preferred over required > address types (link-local and global). Addresses can only be preferred if they actually are candidates for use (e.g, were returned by the DNS, are assigned to an interface, etc.) Individual rules aren't used unless they apply to the addresses being considered. So I don't see any immediate problem if a particular address type isn't used or implemented. Thomas -------------------------------------------------------------------- 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] --------------------------------------------------------------------
