Re: Implementation certification procedure

"Frank Ellermann" <[email protected]> Wed, 20 Aug 2008 17:44:43 +0200
Newsgroups gmane.mail.spam.spf.devel
Organization <URL:http://purl.net/xyzzy>
Message-ID <[email protected]>
Stuart D. Gathman wrote:
 
> What about it guys?  I would "certainly" vote for jSPF as
> "certified", but what about the next implementation that
> comes up?  What is the procedure?

Self-certification is one way to tackle it.  Third party
testing is likely beyond our (voluntary) capabilities.

If a distribution contains its own test suite users could
check that it works also on their platform.  Ideally this
self test can handle any test suite version as input.  

Some weeks somebody used his own "manual" yaml to XML (or
similar) input transformation, and posted his results...

> Did we have a procedure written, but forgot about it?

No, we discussed the problem in early 2007 or late 2006,
but didn't arrive at any convincing ideas, let alone any
written rule.

> Stefano's suggestion to offer a test via the spfquery 
> interface and his live DNS framework is an excellent
> solution.

Indeed, this also covers an open point in the test suite
wish list.  For corner cases like an IDNA <toplabel> it
cannot emulate this effect below a real non-IDNA domain,
but that's as it is.

 Frank