Hello,

Thanks to Scott for his feedback:

> Making DKIM signing MIME aware was specifically rejected during DKIM
> development due to implementation complexity.

I'm afraid I wasn't there, but would like to learn from the past.  Any
references are welcome.


But what exactly do you mean by "implementation complexity"?

 - the need to incorporate MIME-knowledge into an MTA (which one might
argue is not new -- but it is now a requirement for the signing and
verifying MTA, which may have gotten by without until now)

 - the added complexity during signing and verification (I would agree;
but argue that this reflects the complexity of the mail system, and ends
there; it will not grow without bounds)

 - the need for two passes during verification (I am working on that;
recognising an initial bit of text may be better than a rolling checksum
over the entire text)

 - ...?

I think the most important advantage of Lenient DKIM is that it avoids
that a choice made in one place invalidates existing, constructive
things taking place elsewhere.  ARC will not solve that discrepancy; it
imposes one administrator's choice onto others.  To me, that is the most
dire form of complexity (and a reason why people may hold back on
deploying DKIM; look at this email for example, probably being rewritten
to UTF-8 and thereby invalidating my DKIM-Signature made with dkimpy).


Anywhere I repeat things already said, please feel free to point me back
to discussions I've missed.


Thanks,
 -Rick

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

Reply via email to