Re: Implementation certification procedure

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alessandro Vesely wrote:
> I'm thinking about Courier. BTW, it has its own testsuite, see e.g.
> http://www.koders.com/cpp/fidAF68AE4ACFB883D5D3585D26D35899C84E7502AA.aspx?s=md5

That looks a bit too simplistic for a proper test suite.

> IMHO, the best thing would be to have those tuples published on a web
> page. If it were also possible to post the results in order to have a
> response, much like html validation pages, that would be great!
>
> In the latter case, if the submitter authenticates and specifies name
> and version of the implementation, it will be trivial to build a
> database of verified implementations, along with the date and test
> suite version.

Given that mostly just library implementations will be able to be tested
systematically (not on principal grounds, but let's just wait and see how
many non-library implementations, such as MTA plugins, come up with a
command-line testing interface!), I don't see how the effort required to
implement the system you suggest could be justified by the few
implementations that would be using it.  Do you want to do it?

> Stefano Bagnara wrote:
> > I'd happily complete the live dns tester tool but in order to check
> > implementations they have to return an "spfquery" like result, 4
> > lines where the 1st is the result, the 2nd is the explanation, the
> > 4th is the Received-SPF:
>
> What's the 3rd line?

The 2nd line is the authority domain's explanation (via the exp= modifier)
if available, and the local explanation (generated by the SPF client)
otherwise.

The 3rd line is always the local explanation.

> Validating Received-SPF can be slightly more difficult, since order
> and indentation may vary.

Indeed.  This probably isn't something that the official test suite's YAML
file can check.  The best we could do is provide an algorithm (and perhaps
command-line tool) for checking it.

> An empty line to separate one test from the next might suffice;
> however, repeating the tuple as a key is more robust and allows
> results to be reviewed even after new tests have been inserted,
> provided the DNS won't vary.

Please check out the spfd interface, it probably fits:

  http://search.cpan.org/dist/Mail-SPF/sbin/spfd

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkir7ywACgkQwL7PKlBZWjvejACdEfyE6YeO+3WrhrUhF0CIgSjU
xxQAoNTz8ZXjk6dqpE5EIB3c/gsHNoyR
=26UX
-----END PGP SIGNATURE-----