On Thu, 2005-02-10 at 04:01, Tony Finch wrote:
On Wed, 9 Feb 2005, Jim Fenton wrote:
In other words, should the domain publish a gentler policy to allow for
forwarders or should recipients apply a gentler response?
I think the best advice at the moment is to do both of these things. SPF
is probably a reasonable SpamAssassin test, but it isn't accurate enough
to be the sole reason for rejecting a message.
This implies that SPF processing and SPF records will become more
accurate over time just by the virtue of existing. This is bogus, it's
like saying that we want 2+2 to equal 4, but today it equals the
inconvenient 5, but if we just wait long enough, it will equal 4 in six
months. The SPF process checks a fixed set of attributes about a mail
connection against a domain's specified SPF record. Given the same
inputs, the output should be exactly the same every time -- it is only
as accurate as the inputs.
The processing of SPF records and how to specify an SPF record that is
accurate for any given domain is well laid out. Anyone who sets up an
SPF record should make sure that it either reflects how their network
currently operates or how they want it to operate. If they are
specifying how they their network to operate, then they must be
proactive in migrating their network to their ideal setup. These things
don't happen by themselves.
On two occasions I have been asked [by members of Parliament],
'Pray, Mr. Babbage, if you put into the machine wrong figures,
will the right answers come out?' I am not able rightly to
apprehend the kind of confusion of ideas that could provoke such
a question.
-- Charles Babbage
Andy.