On 10/05/2010 04:07 PM, Murray S. Kucherawy wrote:
Hi Rolf,
-----Original Message-----
From: Rolf E. Sonneveld
[mailto:R(_dot_)E(_dot_)Sonneveld(_at_)sonnection(_dot_)nl]
Sent: Tuesday, October 05, 2010 6:27 AM
To: Barry Leiba
Cc: DKIM Mailing List; Murray S. Kucherawy
Subject: Re: [ietf-dkim] Working group last call on
draft-ietf-dkim-implementation-report
Chapter 2: as for DKIM specific definitions used in this document, they
cannot be found in [EMAIL-ARCH]. Please add [DKIM] here as source of
definitions.
The top of chapter 2 contains "[DKIM]" which is the formal reference to that
specification.
Are we looking at the same report? I used
http://tools.ietf.org/html/draft-ietf-dkim-implementation-report which
brings me to version 02 of the document. I see [DKIM] mentioned here and
there, throughout the document, but not in chapter 2? Or do you mean the
page header, which mentions RFC4871 and of course presents the same
referral as [DKIM] gives?
"[EMAIL-ARCH]" is a general tutorial/overview about email architecture,
included to help any newcomers to understand the context of the whole work.
I know [EMAIL-ARCH], it's definitely a good choice to refer in your
document to this architecture RFC.
Par. 3.2 - 3.4: these paragraphs are a bit 'vague', due to terms like
'Some implementations [...]', 'Some test cases [...]', 'at least two
implementations...'. Presumably this has to do with the fact that the
interoperability event took place long time ago?
The language of that section is just a general description of the
participants (the software, specifically) in the interop event and what they
brought to the party. The thrust is actually to show that they were all
diverse, i.e. that a number of different parties wrote code per the RFC and
then tried to send mail to each other to see what happens. It would be
uninteresting to have 20 parties come together all running different
installations of the same package. Showing diversity in an interoperability
report is quite important.
Changing "Some test cases" to "Test cases" seems fine to me.
"At least two implementations" is a specific IETF requirement for an interop
report. I actually am pretty sure all implementations there implemented even
all of the optional stuff, except "z=", but I don't have any specific record
saying that so I'm making only the weaker of the two statements.
OK.
If so, I'd suggest to
include a sentence to describe the difficulty to gather all relevant
information, three years after the event took place. To make these
paragraphs a bit less 'vague' I would like to suggest to include the
original set(s) of test messages and the original interoperability test
plan(s), if still around somewhere.
There wasn't a formal plan. It was basically "send all your tests at all the
participants, then work with each of them to figure out what failed and why".
Each participant brought its own corpus of test messages.
Unfortunately very little of the actual test data and results documentation
still exist, mainly because of a promise among the participants to keep the
results about specific implementations private.
OK, then maybe that's something that needs to be mentioned, to explain
why that part of the report is less exact than for example the data
points about OpenDKIM and AOL.
The public result, appropriately, was the group's evaluation of the
quality of the RFC and the published list of errata.
Apart from these minor remarks, in my view, it's ready to go.
/rolf
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html