I wasn't "there" at the time, but if ----- Original Message ----- From: "Robert Honore" <robert@digi-data.com> To: "Tony Hain" <alh-ietf@tndh.net> Cc: <ietf@ietf.org>; <ipng@sunroof.eng.sun.com> Sent: Tuesday, August 26, 2003 12:40 PM Subject: Re: Solving the right problems ... > Problem is that right now, every time > an application needs such a service, the application programmer must > code his own provisions for it, and none of the propositions I have seen > discussed so far seems to deal with that. couldn't have been a quote about congestion control in 1980, I'm surprised. And now, of course, we accept that transport layers are responsible for congestion control. This makes me think that having something underneath the application take care of multiple interfaces is likely a good idea. I'd also go farther on Robert's point - if we don't think of something fairly quickly, there are likely to be a good number of applications that expect to handle multiple intefaces (with varying degrees of skill) lying around when we think of something later. And this seems to me, to be very analogous to firewalls (we "needed" them because we didn't have end-to-end security, but once we had them, we didn't need end-to-end security nearly as badly), NATs (we "needed" them because we didn't have IPv6, but once we had them, we didn't need IPv6 nearly as badly), and other sources of architectural drift... Spencer