> The "Default Address Selection for IPv6" draft's source > address selection seems to prefer addresses of appropriate > scope over "preferred" or non-deprecated addresses. What is > the reasoning behind that? > > For example, a system has one interface configured with a > deprecated site-local address and a non-deprecated global > address. When communicating with a site-local destination, > the draft specifies that the deprecated site-local source > address would be used instead of the global address.
Because using an address of mismatched scope is more likely to cause interoperability problems. For example I believe there have been IPv6 implementations that refused to communicate when the scopes did not match. The scope of an address is visible to the peer (and hence can cause communication problems), whereas the preferred vs deprecated status of one's address is not visible on the wire. If this scenario really occurred in practice I think it would indicate administrative misconfiguration. I think a host that was forced to use a deprecated address by the source address selection rules should probably log an error or do something to bring the situation to an administrator's attention. Rich -------------------------------------------------------------------- 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] --------------------------------------------------------------------
