[ietf-dkim] Recommend not adding DNS RR(s) into DKIM base doc

[Mark Delany]

On Tue, Mar 28, 2006 at 08:51:59AM +0200, Eliot Lear allegedly wrote:
> Mark,
> 
> My larger concern is this: it's important that we get the RR done in
> time for the base so that we have the query rules done, stating that the
> new RR should be preferred to TXT records.

Agreed. And, er, forgive my propensity to density, but I presume
you're suggesting that a separate doc adds risk because the timing may
not work out.

That concern is clearly valid. The trade-offs are that:

a) the base is kept simpler and more manageable

b) from a deployment perspective, we may be able to get the ball
   rolling earlier on DNS s/w support

c) the DNS RR can progress without all the issues that will arise
   with the base

d) a separate doc has considerable precedent. Eg, RFC821 does not
   describe MX RRs. In fact most RRs as far as I can tell are described
   in isolation from the application protocol.

e) some on this WG think that a new RR is a serious distraction and a
   waste of time. De-coupling a new RR avoids their concerns.


I'm not in a position to judge any of these perspectives and none of
them strike me as compelling, but on the whole, if we diligently work
on an RR spec, which I promise to do, a separate doc seems to have
some merit.


Mark.