Re: [spfbis] Review of draft-ietf-spfbis-experiment-05

[Hector Santos <[e-mail]>]

Scott Kitterman wrote:
> Snip.
> 
> That's all true, but not, I think relevant to the question of record reuse. There was consensus 
> in MARID that reuse of SPF records for Sender ID PRA assessments was technically inappropriate.

Inappropriate in what way was not always agreed.

> I don't think it's correct to apply words about the MARID shutdown to changes that were made later.

As an active participant before marid, during marid and after marid, 
and maybe because my perspective of how all these ideas can work 
together, my assessment was much different. I didn't see precisely 
what the SPF v1.0 group were concern about, but then again at the 
time, I still had a hard time seeing how the PRA would address the #1 
key concern where SPF broke down - path independent.  After all, that 
is what gave birth to the other proposals and it was the MARKETING 
reason presented in the news rags.

Putting aside all subjective politics, the technical problems when CEP 
was presented with its PRA concept:

   - XML format for DNS records,

   - PAYLOAD PRA defeated the high focus with SMTP IP and Envelope 
methods
     to address the accept bounce issue,

The PAYLOAD/PRA concerned was resolved with the SUBMITTER idea.  And I 
believe it was pretty clear we didn't want to add more DNS overhead 
with XML records and processing, it didn't seem this will help make it 
a standard and there was also high interest to consider the new thing 
- a new RR type. Considering, over 80% of the use cases did not 
require the PRA, IMO, it was technically appropriate to consider 
getting rid of the XML format and used a single source language to 
covered both.

The claim that there was potential for errors seems to be based on the 
idea that implementators and publishers would do it wrong - not that 
the protocols were incorrect because they were both technically 
distinct with a tag.

I didn't get that part of it, and FWIW, I felt that it could be 
integrated smoothly without problems.  Waste? Perhaps. Lower use 
cases? Perhaps.   But to me, it was more did PRA help (not cure) with 
the SPFv1.0 path problem that was the #1 criticism it had.

I just feel if the SPFBIS focus was on the integrated experiment 
question which would include resolving the DNS record conflict claims, 
then there would be a different set of interest.

-- 
Hector Santos, CTO
http://www.santronics.com
http://hector.wildcatblog.com