Re: [apps-discuss] Updating the status of SPF

[Scott Kitterman <[e-mail]>]

On Thursday, August 11, 2011 05:36:26 PM Andrew Sullivan wrote:
> On Thu, Aug 11, 2011 at 11:19:06PM +0200, Frank Ellermann wrote:
> > If you insist on it the WG should IMHO still be free to say whatever
> > it likes to say about the type 99 DNS SPF RR; maybe kill it for good.
> 
> The plain fact is that once a DNS RRTYPE is allocated, it can't be
> "killed".  You can deprecate its use and tell new software not to look
> it up.  But the code point will remain assigned, I predict, until we
> stop using the DNS (which, at the rate we are adding lard to spackle
> to baling-wire, might be any day now, but that's a digression for
> another list).
> 
> For the little it us no doubt worth, it is the "TXT is now used
> everywhere, and so we're not going to deprecate that even in new
> versions of the protocol, pppptttthhht" argument that annoys people
> who see the problems with TXT records whenever they're overloaded in
> yet another protocol.  In effect, once you adopt TXT, you're going to
> use it forever.  This same attitude is why we still have A-record MX
> fallback all these years later.
> 
> But I do agree that, if people are wedded to using their bad idea
> forever, one isn't going to change their mind, and it is silly to have
> two mechanisms for achieving the same goal one of which is never used
> (particularly if it causes additional DNS load).

I feel strongly that the 4408bis effort we are discussing now should be a 
minimal/backwards compatible_unless_it_is_impossible update.  Once that is 
complete there is a backlog of incompatible design ideas for simplifying and 
improving SPF.  In that context I think moving to Type SPF only makes a lot of 
sense and should be considered.  I don't want to get that mixed in with the 
current effort.

I, for one, am certainly not wedded to the idea of using TXT forever.

Scott K