On Fri 05/Aug/2022 19:18:23 +0200 Scott Kitterman wrote:
On August 5, 2022 9:15:32 AM UTC, Alessandro Vesely <ves...@tana.it> wrote:
On Thu 04/Aug/2022 22:50:20 +0200 John R. Levine wrote:
I think that Ale's expression that he had difficulty understanding the
description of the tree walk as written is a strong sign we still need to
improve the language. Of the people involved in this specific discussion, as
far as I know, he's the only one who's first language isn't English. That's
true for most of the people in the world, so I don't think we can call it quite
done.
I took a look at the description ot the tree walk in section 4.6 and I didn't
see anything to change. I suppose we could add a sentence at the top of the
section reiterating some of the discussion in 4.4 that you do tree walks on two
domains to see if they are in relaxed alignment, or that you do a tree walk on
the RFC5322.From domain to find its policy domain, but those seem marginal.
What Ale seemed to be asking for was something saying these aren't zone cuts
and it's not the PSL. I don't think that would be a good idea. The list of
things we are not doing is unlimited, and mentioning some of them just leads to
questions about if you're not doing them, why did you mention them?
I might have messed up that introductory paragraph several times, but that's
not the main point. *Section 4.8* mixes the definitions of Organizational
Domain and relaxed alignment. That way, it is difficult to guess which
precondition applies to which definition. I think native English speakers may
misunderstand it too.
Please quote the exact text in Section 4.8 in the proposed text that you
believe provides a definition for relaxed alignment. I am unable to see it.
For example:
This means that a DNS Tree Walk to discover an Organizational Domain
will start at once of the following locations.
Organizational Domain, albeit in a DMARC sense, is a generic concept.
Someone, typically a developer, may want to determine it starting from an
input domain, without having to know in what location the domain was found.
IOW, to say that there is no need to check alignment if no DMARC record was
found is different from saying that the Organizational Domain cannot be
determined in the absence of a DMARC record.
In particular, the condition that a tree walk must have met a DMARC record to be eligible
for org domain determination —that is, org domain cannot be determined unless a DMARC
record is found— is not evident. It is expressed only implicitly at the beginning. Then
the paragraph "To discover the Organizational Domain for a domain, [...]" can
be interpreted as the start of the org domain discovery algorithm definition, in the
middle of the alignment discussion, dropping the preconditions stated for alignment.
The first part of 4.8 defines the input process of doing a tree walk to get a
set of DNS data to use as an input for organizational domain determination.
The second part (as proposed) describes how to use that data to determine an
organizational domain. In my experience, describing inputs to a process is
both normal and necessary.
No, the input is a tree walk starting from a domain. A pre-condition is
that one or more DMARC records have been found in the tree walk. This is
implied, not said.
The juice of my proposal is to define the org domain discovery algorithm in its
own section, and discuss relaxed alignment elsewhere.
I think that's exactly what the current text does.
No, despite the title, Section 4.8 describes checking alignment.
Best
Ale
--
_______________________________________________
dmarc mailing list
dmarc@ietf.org
https://www.ietf.org/mailman/listinfo/dmarc
