John Mears wrote in
 <lo6p265mb69883c6f3219283ce1be2b76e8...@lo6p265mb6988.gbrp265.prod.OUTLO\
 OK.COM>:
 ...
 |>It appears that Richard Clayton  <[email protected]> said:
 |>>>> 4  repeat until you get back to the original state of the message or a
 |>>>>     "z" recipe tells you that you need to unconditionally trust
 |>>>>     someone...  they presumably have placed an Authentication-Results
 |>>>>     header field into the message and you have to use that for \
 |>>>>     your DMARC
 |>>>>     and/or reputation calculations.
 |>>>
 |>>>I guess in this case A-R has to be signed? What handling is suggested
 |>>>in such a "z" case if you _don't_ trust the modifier?
 |>>
 |>>if you don't trust someone who modifies a message then I think you
 |>>should refuse to accept the message.
 |
 |>The idea of the "z" tag is that it's for filtering front ends like \
 |>Proofpoint and Mimecast that rewrite URLs and strip attachments. They \
 |>would only be doing that to mail sent to recipients who are paying \
 |>them to do it.
 |
 |>This does mean that remailing such a message back out won't work with \
 |>DKIM2 but I haven't seen a lot of mail like that.  The rewrites tend \
 |>to appear in replies or forwards that are new messages so they're OK.
 |
 |I speak as someone designing/implementing/running such a filtering \
 |system. Yes, those kind of modifications (URL rewrites, attachment \
 |removal) are generally done on inbound processing shortly before delivery \
 |to the recipient's mail system. It certainly does happen in some situati\
 |ons that the recipient's mail system can forward the original email, \
 |rather than creating a new email including the original content. Question: \

 |I guess it is up to the filtering intermediary's discretion whether \
 |it uses z= rather than spell out recipes for the modifications made, \
 |accepting the trade offs? For example the filtering intermediary might \
 |use z= if the original has known malicious content that it wants to \
 |remove all trace of - versus an email that it can establish is established \
 |not malicious, in which case it would be useful for the forwarded recipi\
 |ent to have the recipes for reconstruction?

Very interesting question i was wondering why noone else asked
them (in general) .. in public and on the ML.

For the ACDC draft i speak and say that if you destruct the
original message never to be seen again, you create a new
message: the original variant cannot be brought back.

One could add yet another flag to signal this special condition,
say "T"ransplanted.  However, one could also throw away all elder
DKIM (and DKIX, thus) header fields, because they do not mean a
thing, they would cause "z" (non-reversibly modified).  Using "T"
would shine a more smooth light onto any "z" that happens later
on thus, and only.

The same is true if "a user" throws away the DKIM-Store: header
field of the ACDC approach, ie what carries state from ingress
to egress.  (If it is done like that software-wise, "surely
is" for any non-integrated thing that uses external software
for processing.)  As there is no way to recover any former state
(except "status" through the possible "I"ngress signature, unless
that is also removed) without it.

Maybe "T" is a good thing, and explicitly stating that in such
cases any elder -DC header field should be removed as it became
useless would then also make sense.  And the "sequence" would not
need to be reset to "0", in that case, even though any elder DKIM
signature is "hanging in the air".

It is solely a matter of trust to the hop which "T"s whether
anything elder is not pure halluzination.  That is a lot of
trust.  I remember having perceived PGP signing parties with
passport request, announced by computer magazines weeks and
months in advance.  Paying money for a service seems to go in the
same direction.

So "T" for "Tranplantation" as well as for "(total) Trust".

Having said all that i have never consciously seen things
removed except for policy reasons (images, HTML MIME parts,
(shell) scripts and executables) -- and those, except for their
sheer size (and possible danger), could very well be part of
a reversibility plan.  I wonder what needs to be hard cut like
that, and [.]

Btw your message made me very interested, so, for whatever
interesting reason, i actually did reread the current drafts as
mentioned on the DKIM WG internet side, .. and it turns out there
is no z=?  (It seems to have existed in the past, there.)

[.] it seems to me plain rejection was not only my thought.

 |New to this thread so apologies if I am reopening old wounds.

Well the ACDC isn't it for the WG, but i thank you nonetheless,
and it seems i should not only talk less but also iterate the
draft once more, i have seen in the WG draft they have and i miss
IANA registry requests for the new header fields, too.  No hurry.

Legal and such issues i would have expected to play a much bigger
role.  (It is different if you are only a glancing reader, or
actively involved in something.  Ie, maybe it has always been like
that.)

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)

_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to