Re: [v6ops] I-D Action: draft-vyncke-v6ops-happy-eyeballs-cookie-00.txt

[Simon Perreault <[e-mail]>]

On Thu, Oct 30, 2014 at 8:16 AM, Bjoern A. Zeeb <
[e-mail]> wrote:

> > On Oct 29, 2014, at 5:28 PM, Bjoern A. Zeeb <
> [e-mail]> wrote:
> >> I would phrase that differently as “I want predictable, deterministic
> behaviour and not magic”.
> >
> > Then you want POTS, not a global packet-switched network.
> >
> > :)
>
> Damn that thing still works after so many decades and is so reliable
> indeed;  now SIP is taking over.
>
> I was more thinking in terms of applications (or more general end-node
> behaviour), but I wouldn’t mind more simple, deterministic protocol
> behaviour either;  less can be more;  feature-itis is a disease of its
> own.  And, personally I guess, if we had more working (deployed) code
> before we’d have RFCs published on specifications, we’d have a lot less OPs
> (or ooops) things to write down and deal with and a lot less crap and a lot
> less drafts making it to RFCs;-)


Come on Bjoern, you're too young for change resistance syndrome! ;)

We need more and more to think of source IP addresses as ephemeral things.
Just like you don't care which ephemeral source port gets picked by the OS
when you call connect(), you shouldn't care about which source address is
picked when calling connect_by_name() or whatever other HE-like API out
there.

Simon