> > While I don't have any data on this, there's the issue I pointed out in
> > other e-mail about dial-up users. No system I know of implements the
> > "quiet time" concept when it logs in to the Internet by means of PPP or
> > whatever.
> That would be a bug - a big one. Easy to fix, and required.
No thanks. When I dial in using PPP over my cell phone from an
airport lounge where I'm catching a connecting flight, I don't want to
have to sit there for 3 or 5 minutes while my network stack enforces
some sort of "quiet time". I want to grab my e-mail now.
I'm not so worried about old packets causing trouble. I'm tunnelling
things over ssh. And my TCP has the timestamps option enabled. So I'm
safe enough (from corrupting my e-mail fetch) without the quiet time.
It may or may not be out of spec, but ignoring the quiet time after
establishing a PPP connection is the right thing to do.
The other methods of decreasing the likelyhood of a stale packet causing
trouble are to be encouraged. (E.g. use of timestamps option, use of
random port numbers, randomized initial sequence numbers, IPSEC's ESP,
tcpm mailing list