spf-discuss
[Top] [All Lists]

Moving forward

2004-09-29 15:35:03
Hi all -

I don't post much here, so many of you may not know me: I'm your humble standards editor and co-author with Meng.

We (Meng and I), as authors of drafts that were part of the MARID working group, have been asked to "put forward their documents as non-working group submissions for Experimental RFC status". Experimental RFC status is granted to a draft that "is part of some research or development effort" and "is published for the general information of the Internet technical community and as an archival record of the work" (to quote RFC 2026).

The question before me is, what do I include in such a submitted draft?

If we were a group with a clear, though evolving, design, and running, interoperable code, then this would be the perfect time to document that which has been stable and working, bundle it into a draft, and submit it.

But we are not that. Discussion on this list still makes it clear that even within the SPF faithful, there are vast disagreements on what should be checked, what order, what the meaning is, and the even the details of the technical form. Indeed, even my current thoughts on SPF (which I'll express to this list at a later time) are now tending to some more radical changes (mostly in the form of excising as much as possible!)

At the core I see there are set of ideas, that one might label "SPF Classic" that seem to endure. But in active discussion and development I see many different splinters of this: Meng's call for a new form of Unified SPF, various combinations with and with out CSV, my own thoughts toward a stripped down SPF, and even Microsoft's call for submitting the dual-identity version of Sender ID. Alas, these changes don't seem to me to be tweaks of the last ten percent -- they are often wholesale changes in approach.

If I'm to write up what should be put forth for experimental status, then we need to have some cohesion on what it is that we agree on for a core. The only thing I see is the SPF Classic ("v=spf1") that is deployed in thousands of domains and implemented. Do we have enough agreement to promote that? If we should be promoting something else, can we show enough rallying around some other approach to claim that it is meaningful to document it now so experimentation with interoperability can begin?

        - Mark

Mark Lentczner
http://www.ozonehouse.com/mark/
markl(_at_)glyphic(_dot_)com


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