[e2e] Discrete IP - retake

[Jon Crowcroft]

There is such a thing as too many possibilities - look at
protocols like the Session Initiation Protocoll (aother SIP:)
which is just too much of an xmas tree temptation

IPv6 is similar -whereas Steve's Simple IP didn't offer such a
huge temptation to put everything and the kitechen sink - it was
just IP with more bits, but all the same hacks you have with IPv4
address space, but just the luxuery of 2^32 times as much space
(or 2^32 IPv4 address spaces if you wanna think of it like that)

yes, you are right, novell netware had enough of a footprint out
there that the compromise/committee decision to go for 128 bits
looked attractive for the same wrong reason that adopting CLNP and
NSAPS looked attractive earlier (because DECNET already used it,
so you got magic early adopters for free:)

imho, a bad decision. (but this is a bit of hindsight and also, at
the time, being heavily involved in the ipng requirements capture,
and in the PIP effort towards ipng, I had an axe to grind - but i
don't think the result did any favours to any particular group in
the end....

the id/locator stuff might still have happened with SIP (and
certainly would have with PIP, as it was a "feature") so who knows
if we'd have actualy got there sooner...

meanwhile, when I was a kid, I remember visiting our next door
neighbours for sleepovers, and ther parents would read us the
hobbit. (this is in about 1961 i guess) - at a certain point, my
three friends, three brothers called Tom, Bert and William
(a genuine serendiptous coincidence, I learned later)
would exclaim with delight, "that's us - this is all about us!!"

I assume you know which bit I am talking about

of course, they came to a sticky (well, rocky) end (the characters
in the hobbit, not my friends:-).

I'm looking forward to the movie very much

In missive <20120918181654.B2BF118C0D3 at mercury.lcs.mit.edu>, Noel Chiappa t
yped:

 >>    > From: Jon Crowcroft <jon.crowcroft at cl.cam.ac.uk>
 >>
 >>    >> How would SIP have been any easier-to/better-at actually being
 >>    >> deployed than IPv6 
 >>
 >>    > No 8+8, loc/id split endless timesink debate would have happened with
 >>    > SIP as there weren't enough bits
 >>
 >>That had no effect on the deployability, though. To the extent that discussion
 >>had any effect on IPv6 at all, it was purely on the schedule - and I'm not
 >>even convinced it had much effect on that.
 >>
 >>BTW, you probably recall this, but for those who weren't there, the switch
 >>from 8 byte address size to 16 bytes had nothing to do with location/identity
 >>separation, it was down to wanting to support Netware (I think it was - some
 >>XNS derivative, anyway) addresses in IPv6. (The change was made at a meeting
 >>in Chicago - if anyone care, I can probably dig up a reference.) 8+8 and all
 >>the other location/identity separation schemes came later, once people
 >>realized there were enough bits there.
 >>
 >>	Noel

 cheers

   jon