> I think this discussion is mostly over (for now), so I will respond > to just one point in your message, and leave the rest for > perhaps some other time... > > | okay. but there's a fairly fundamental conflict here between this > | use of LL (in a network that will never be connected to the internet) > | and the use of LL on a network that is only temporarily disconnected > | from the internet. apps need to know which kind of address to use. > > Hmm - I think apps (&/or the system, or something) need to be able to > do something here (if they're the kind that survive long enough for > changes to occur). Things like MTAs, DNS servers, ... all do already > (or do if they're worth using). > > That is, it is entirely likely that a node will power up with no net > connected (hence no global addresses - v4 or v6), and apps will > start running. Later it connects to a net, and globals appear. > Later still, the net goes away again, and the globals expire. > When the net comes back again, the global addresses that reappear > may be the same as before, or they might be different. > > All this happens all the time - as things are now (regardless of whether > scoped addresses exist or not).
indeed. > Apart from simply refusing to work if a global addr isn't available > (as distinct from being able to contact some remote site, which is > kind of inevitable) apps really need to be able to deal with things > like this. agreed. apps really do need a way to deal with things like this. unfortunately, they don't have a good way of dealing with things like this, and it's not at all clear that placing the responsibility with the apps to deal with things like this is architecturally sane. > Like it or not, the world has changed in the past 20 years indeed it has (and along with it the net), and this is the entire point. the net has changed in such a way that it doesn't support some kinds of applications very well. at the same time, we have an even greater desire to support those kinds of applications. > - your > typical internet node is no longer a half ton (or more) beast that > runs forever, and never turns off or gets disconnected. Apps, and > app designers just have to adapt - regardless if this means doing > more work. it's not at all clear that the problem can be solved effectively by placing the burden on the apps - in fact it's clear that the problem *cannot* be solved effectively merely by expecting apps to do "more work". like you say, the net has changed in the past 20 years, and the days in which somebody could make a change to one aspect of the net and claim that whatever difficulties those changes caused are somebody else's problem, are long gone. 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] --------------------------------------------------------------------
