On Sunday, March 20, 2022 6:47:07 PM EDT Barry Leiba wrote:
> > No, that is completely wrong.  Please review the previous thousand
> > messages about why we are switching to a tree walk.
> 
> Rather than being snarky about reviewing the previous thousand
> messages, perhaps you could write up a one-paragraph summary of the
> result?  Hhere's why:  If it took us a thousand messages, literally or
> metaphorically, and if some are still showing a lack of clarity about
> why we got here, then (1) a brief reminder posted here wouldn't hurt
> and (2) readers in the future are going to want to know as well, and
> having something we can put in the document to explain it will be
> really useful.
> 
> Then the text we settle on can go into the draft and it will be
> clearly documented so that others -- future implementors, the IESG
> during the approval process, whoever -- will know the story and will
> have fewer questions.

Generically, I think that's quite reasonable and I'll give it a shot.  
Specifically, I am also feeling some frustration about this.  My impression is 
that whenever this topic gets mentioned, there are certain participants 
attempting to rehash the argument over again and unwilling or unable to accept 
the arguments that have been made.  I didn't look back and see who wrote this 
or who it was written to, so I have no idea about the conversation previous to 
this on the thread.

Working from memory (so chime in if I miss something):

The PSL was created to deal with the lack of a clear way to distinguish DNS 
boundaries related to web technologies.  It's purpose is unrelated to email.  
While it served reasonably well for the initial DMARC deployment, it has a 
number of limitations that lead it to not being an optimal choice for 
determining DMARC Organizational Domains:

1.  It is by design, incomplete.  Due to lack of resources, the PSL 
maintainers have stated that not all domains can be included.

2.  Due to the differing purpose of the PSL, there are cases where the correct 
data for the PSL is not the correct data for DMARC (see the recent message 
from John Levine with examples).

3.  PSL is maintained by a single private organization with limited external 
oversight and no documented configuration management process.  It could 
disappear tomorrow or be completely changed if the people running it decided 
to do so.  A protocol based on an unstable reference such as this is not 
suitable for standardization by the IETF.

The alternative Tree Walk mechanism that the working group is currently 
finishing work on has none of these disadvantages.  We believe it will produce 
more correct results and be more maintainable than continued used of the PSL.  
We also believe that, in practice, it is backward compatible with existing 
usage.

To the extent we have found differences between the results with PSL versus 
Tree Walk, the Tree Walk results have all been more correct.

My predication is we will now get to redo the argument about this, just like 
we do every time it comes up.

Scott K


_______________________________________________
dmarc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dmarc

Reply via email to