>> I'd disagree, actually.  Not all uses of the resolver are to try to
>> initiate communication with, and *definitely* not all uses of the
>> resolver are to try to initiate communication with from the current
>> host more-or-less immediately after the query.
> It's not _that_ bad... you'd  have to ask the resolver, load a kernel
> module that implements the protocol, then initiate the connection to
> see the problem.

Well, to see the problem with a connection.

> But there's another scenario:

> You could ask the resolver, then pass the address to a network stack
> implemented in userland using, say, bpf.

Yes.  Or you could, for example, ask the resolver, then print out the
result for a human to do you know not what with.  Or you could ask the
resolver and send the results to another machine which is responsible
for setting up the connection.  (Why can't it ask the resolver itself?
Perhaps because it does a lot of connections and wants to offload the
resolver work onto its clients.)  Or, or, or...or lots of things.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B