> There is no difference in a SL vs. global prefix in the process of > address creation in the stack. The real difference is that the simple > device frequently is concerned about memory consumption, and that is > minimized when it only has to support one SL & LL. The light switch > could have a simple policy that the only prefix in the RA that it looks > for is SL, which would minimize the management and make it more > plug-n-play.
such a policy would also make it dysfunctional. it's completely bogus for such devices to impose constraints on how a network is designed. if a device is only going to recognize one RA prefix, it should probably be the shortest prefix that it sees. > I am not claiming I know of devices that look like this, but the > opportunity exists the way things are currently defined. If there are > multi-party applications that can't deal with scope boundaries, they are > aware of that limitation and should be provided a mechanism to tell the > stack SL is not an option for this socket. oh right, so we need to configure each application to be aware of network topology to know when to use and when not to use SL. > My real problem is that the thread is about preventing or restricting > any use of the address format, simply because there is one valid case > where it is not a good choice. so far there is at most one case where it is a good choice - providing a way for local connections to survive renumbering. and there are almost certainly better ways to solve that problem - defining explicit values for minimum time between renumberings and maximum time that a connection should be expected to stay up would be a good start, and they wouldn't be specific to local connections. > The fact that there are other cases which are valid is being ignored. we're still waiting to hear about them. > If someone wants to write a BCP that says SL > should not be used for multi-party apps, fine, but this doc should not > go further because other uses do not share the problem. no, it's the other way around. if someone wants to impose the constraint that large portions of the network should only support client-server apps then they need to bear the burden of convincing the rest of us as to why it's in the best interests of the Internet to have such a constraint. Keith -------------------------------------------------------------------- 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] --------------------------------------------------------------------
