Re: [dmarc-ietf] Discussion

I feel (as a receiver of reports) as though another reason to have these as a single report is so that the receiver of the report is able to more easily correlate the data/times.  I should be able to believe that all reports (without a specified ri) will be 0000-2359, though things happen, and systems break down.

--
Alex Brotman
Sr. Engineer, Anti-Abuse & Messaging Policy
Comcast

From: dmarc <dmarc-bounces@ietf.org> On Behalf Of Marc Bradshaw
Sent: Monday, December 7, 2020 5:17 PM
To: DMARC IETF <dmarc@ietf.org>
Subject: Re: [dmarc-ietf] Discussion - ARC/Extensible Reporting (Ticket #56)

There is value in including these in one report, especially where the extended property being reported on influences how DMARC is evaluated.

Using ARC as the example, when a p=reject policy was overriden by the receiver because of an ARC evaluation, when that data is included in the report the indirect mail flow can be clearly seen, which would not be the case if ARC were included as a separate report.

This could be included in an <extension> element, or could be included directly in the auth_results section of the report, based on an assumption that the things being reported SHOULD have an authentication results entry in processed mail.

<auth_results>
<dkim>...</dkim>
<spf>...</spf>
<arc> (as defined in arc standard) </arc>
<foo> (as defined in foo standard) </foo>
</auth_results>

These would be defined in a registry with reference to each standard, and parsers would ignore any sections they did not understand.

On Fri, 4 Dec 2020, at 6:43 AM, Alessandro Vesely wrote:
On Thu 03/Dec/2020 19:54:51 +0100 Brotman, Alex wrote:
>
>> -----Original Message-----
>> From: dmarc <dmarc-bounces@ietf.org<mailto:dmarc-bounces@ietf.org>> On Behalf Of Alessandro Vesely
>> Sent: Thursday, December 3, 2020 6:24 AM
>> To: dmarc@ietf.org<mailto:dmarc@ietf.org>
>> Subject: Re: [dmarc-ietf] Discussion - ARC/Extensible Reporting (Ticket #56)
>>
>> On Wed 02/Dec/2020 20:46:54 +0100 Brotman, Alex wrote:
>>>
>>> While this ticket/enhancement specifically mentions ARC, I could
>>> perhaps see the usefulness in other places.  It seems like it would be
>>> more beneficial to create a method by which other documents could
>>> provide XML- based "extensions" to the report.  This would allow
>>> mechanisms relying on DMARC to independently define their reporting
>>> schema to be included in DMARC aggregate reports.  Alternately, we could
>>> focus specifically on ARC, and work to include that in the base XML.
>>> This means that any later reporting requirements could again require
>>> changes to the core drafts.>>>
>>
>>
>> Another possibility is for ARC to define its own report format.  Hijacking
>> rua= targets to send a different kind of report should be allowed.
>> Otherwise, we could define a new tag, e.g. rue= (e for Extension).>>
>> In either case, as we're introduce variations in aggregate report content,
>> we have to devise a method for determining what version/kind of report is
>> attached to a given message.>>
>
> We could add an element called "<extensions>", and we allow ARC or whatever
> it may be to exist under that element.  The Aggregate Reporting document
> needs to specify that any extensions are expected to be proper XML, and if
> there are no extensions, an empty element is sufficient.  We could create a
> bit more structure as a requirement if we wanted:>
> <extensions>
>    <extension name="arc" standard="ARC_DMARC_REPORTING_EXTENSION_DEFINITION">
>      ... (as defined in referenced standard)
>    </extension>
> </extensions>
>
> If a report parser doesn't know what ARC is (or any of the extensions), it could skip the processing.  I do understand this means that <extensions> element may break existing parsers, even when empty, though, I expect many of the things we're proposing may fracture the expected XML.

We can do a better job at producing aggregate reports with an automatically verifiable content.  For example, prepending stuff like this:

<?xml version="1.0" encoding="UTF-8"?>
<dmarc:feedback xmlns:xs="http://www.w3.org/2001/XMLSchema-instance<https://urldefense.com/v3/__http:/www.w3.org/2001/XMLSchema-instance__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCW5mqmCE$>"
xmlns:dmarc="http://dmarc.org/dmarc-xml/0.1<https://urldefense.com/v3/__http:/dmarc.org/dmarc-xml/0.1__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCX_7HCv_$>"
xs:schemaLocation="http://dmarc.org/dmarc-xml/0.1<https://urldefense.com/v3/__http:/dmarc.org/dmarc-xml/0.1__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCX_7HCv_$> rua.xsd">
<report_metadata>
        ...

(Perhaps something better than "http://dmarc.org/dmarc-xml/0.1<https://urldefense.com/v3/__http:/dmarc.org/dmarc-xml/0.1__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCX_7HCv_$>" for the version)

But then, would the <extensions> imply dmarc-xml grammar should be upgraded every time ARC (or whatever) is upgraded?

Separate reports sounds simpler.

Best
Ale
--

_______________________________________________
dmarc mailing list
dmarc@ietf.org<mailto:dmarc@ietf.org>
https://www.ietf.org/mailman/listinfo/dmarc<https://urldefense.com/v3/__https:/www.ietf.org/mailman/listinfo/dmarc__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCcp8JAZr$>

--
[https://secure.gravatar.com/avatar/b214a020f4eb135ce2a6901d7540bdb1?s=44&d=404]

  Marc Bradshaw
  marcbradshaw.net<https://urldefense.com/v3/__http:/marcbradshaw.net/__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCcwnsT8Z$> | @marcbradshaw<https://urldefense.com/v3/__https:/twitter.com/marcbradshaw__;!!CQl3mcHX2A!W_xDXVx707gc61M5TAh1t1CL61wpj57BHndbtFxULnxLzggrXnZfXLDP5T6PCZsg4P4v$>