Re: [spfbis] Proof of non-deployment [[e-mail]: Cron <root@primary> /usr/local/libexec/spf-txt.sh]

[Andrew Sullivan <[e-mail]>]

On Tue, Oct 07, 2014 at 09:15:49AM -0700, Murray S. Kucherawy wrote:
> 
> Given the survey work recorded in RFC6686, I would be curious to hear
> people's theories explaining the substantial uptick of TYPE99 queries in
> the last two years while the industry has actually gone in the opposite
> direction.  Of course, that presumes there's ample breadth and no bias in
> the traffic seen by this one resolver.

We had methodological issues at the time of RFC 6686, too, so I think
we ought to be careful about speculating too much.

But I seem to recall someone telling us, right around the time of
6686, that some new SPF checking code had just shipped that used
TYPE99 as the default.  This news came too late in the publication
cycle to change the text.  But it shouldn't be surprising that we see
this now.

None of that matters, however, because the original RFC had that
interoperability bug in it, and the only reasonable thing to do was to
pick one type to look up.  I'm unhappy about using TXT, too, and
registered that unhappiness at the time.  But the deployed code is a
strong argument.

A

-- 
Andrew Sullivan
[e-mail]