Re: interop problems with getaddrinfo() address selection

[der Mouse <[e-mail]>]

> The examples you listed, based on different kinds of connectivity,
> are reasonable; what it not is de-randomizing equivalent classes of
> address.

But whose definition of "equivalent" applies?  From a client point of
view, close addresses are not equivalent to distant addresses - the
client wants a nearby server, not unreasonably - making sorting by
match length at least moderately reasonable.  (Not entirely; routing
topology doesn't necessarily parallel matching prefix length.  But,
absent something like AS path information, it's probably the best
available approximation.)

>> As for DNS randomization, it's never been reasonable to expect an
>> application to try each of several addresses in the order returned
>> by DNS.
> A lot of sites are going to be rather upset if wide deployment of RFC
> 3484 suddenly breaks their service architecture.

Any architecture that depends on the DNS preserving record ordering
(never mind depending on clients doing anything in particular with the
order the DNS happens to give them) is already broken.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       [e-mail]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B