[Top] [All Lists]

RE: Update on Broken SPF Records

2005-08-14 07:33:00
...... Original Message .......
On Sun, 14 Aug 2005 01:32:58 -0500 "Herb Martin" 
-----Original Message-----
From: Scott Kitterman [mailto:spf2(_at_)kitterman(_dot_)com] 

Herb asked:
Is there an official one that is both RFC compliant and RFC strict?

SPF is dangerous to implement (due to the way some people use the 
results) if there is no way to determine the correctness.

The one I did is the closest thing to it that I know of. If 
the are any ways it doesn't align to the current draft, 
please let me know and I'll fix it.


Well, my "main" triplet(s) work:  IP/Helo/User
but the "too many lookups" is still pretty worthless
without some debug information.

And things like "mx" -- does that count as TWO lookups?
(One for MX list, and one for a particular Server, or does
each MX server get counted?  You could easily get to 5 or
more right there.)

For example:  Hotmail.com has 16 MX records alone -- which
would make a minimum of 17 lookups if you count the MX
request and had to check all of them to find the last one.

If there is such a rule as the "10 lookups", it really should
never apply EXCEPT for a particular SPF test.

This 10 lookup rule may destroy what value is left in SPF
-- which has already taken a beating in public perception
due largely to the misuse of SPF by administrators who
didn't understand it's real value.

(I was a strong supporter of SPF until that was brought
to my attention -- now I am re-evaluating that support.)

MX counts as one.  All that counts are the mechanisms that are listed in 
your record or the ones you include: or redirect=.

I took a quick look at your record.  I expect it can be simplified and am 
willing to help you with that.  Don't give up yet.  BTW, I don't currently 
see an SPF record at spf.learnquick.com.  That will be a problem.

My initial focus on the validator was to make sure only compliant records 
would pass.  I agree it's not an ideal troubleshooting tool.  I intend to 
make it better.

Scott K

<Prev in Thread] Current Thread [Next in Thread>