Re: [dnsext] need new flag bit in EDNS, "do me no favours" (DMNF)

[Paul Vixie <[e-mail]>]

> From: Roy Arends <[e-mail]>
> Date: Sun, 24 Oct 2010 23:01:14 +0000
> 
> > opin?  i can write a short i-d on it before beijing.
> 
> The end-game will be applications doing their own resolving. Real
> control.  No third party dependencies. No favors to ask.

since the end game for serious apps is stub validation, i tend to agree.
however, for non-serious apps i think ISP-level or even ASP-level rdns,
if an rdns operator wants to offer opt-out on web error redirection, i
think there ought to be an in-band way for stubs to opt out.

it's likely that the BIND resolver would implement DMNF defaulted to "on"
but allow it to be set to "off" in /etc/resolv.conf.  and it's likely
that opendns and possibly other ASP-level rdns operators would respect
this setting as part of their opt-out strategy and implementation.

and depending on the interpretation -- in other words, this would be left
loose deliberately in the standard for DMNF -- an rdns server who normally
hides AAAA RRsets from queries coming in on a non-IPv6 transport, would
turn this off in the presence of DMNF=1.  so there are other applications.