DKIM -base Open Issues Eric Allman IETF 65 March 20, 2006 carryover: draft-allman-dkim-base-01.txt - Should we have an r= tag in either the signature or key record 1183 lear(_at_)ofcourseimright(_dot_)com OPEN no thread? There is a thread on making r= localpart only (from Mark D) carryover: Develop plan for transition of multiple crypto algs (a=) 1184 lear(_at_)ofcourseimright(_dot_)com OPEN not much discussion of how to transition, though not much disagreement either 3/9: “Not much discussion; not much disagreement” http://mipassoc.org/pipermail/ietf/dkim/2006q1/002414.html carryover: draft-allman-dkim-base-01.txt Transition sha-1 to sha-256 1185 lear(_at_)ofcourseimright(_dot_)com OPEN not quite closed on the actual exact wording [I think we had converged on MUST accept either, SHOULD generate sha-256] http://mipassoc.org/pipermail/ietf-dkim/2006q1/002414.html base spec: instead of signing the message, sign the hash 1193 lear(_at_)ofcourseimright(_dot_)com OPEN no (recent) thread Summary: Hash the body, store that in header, hash and sign the header Hash could be in DKIM-Signature or another header field base spec: whitespace in signature? 1194 not sure if this is the right thread OPEN “Need to use appropriate folding rules for signature line (CFWS, et al)” http://mipassoc.org/pipermail/ietf/dkim/2006q1/002464.html (Message not found) draft-ietf-dkim-base-00 - 3.4.6 Example (Canonicalization) 1195 hsantos(_at_)santronics(_dot_)com OPEN no discussion “1) Please note "relaxes" typo in 3.4.6 example: "Assuming a "c=relaxes/relaxed" canonicalization algorithm, a message reading:” [Fixed] “2) Consider adding more examples to illustrate our possible algorithms and combinations.” http://mipassoc.org/pipermail/ietf/dkim/2006q1/002148.html Base: Upgrade indication and protection against downgrade attacks 1196 MarkD+dkim(_at_)yahoo-inc(_dot_)com OPEN lots of discussion, no clear closure Summary: add tag in selector record indicating lowest algorithm that will ever be used for signing http://mipassoc.org/pipermail/ietf/dkim/2006q1/002163.html MUST vs SHOULD in Verifier Actions section (-base) 1200 eric(_at_)sendmail(_dot_)com OPEN “There are several places in the Verifier Actions section of draft-ietf-dkim-base-00 that say that a verifier MUST ignore bad or malformed signatures. This is really a local policy question, and we have been trying to stay out of that. Shall we change these to SHOULDs, or even just change these to read something like "Bad or malformed signatures MAY be ignored. This is a local policy decision and beyond the scope of this document."?” change the syntax from SPF compat to human compat 1201 MarkD+dkim(_at_)yahoo-inc(_dot_)com OPEN See 1217: SSP: should we drop the cryptic o=. syntax for something a little more readable? http://mipassoc.org/pipermail/ietf-dkim/2006q1/002219.html Really not appropriate for this session — SSP-specific
 extendable RR records? 1203 tony(_at_)att(_dot_)com ACCEPT the title of this issue is misleading, its really about extra options to be specified in a DKIM TXT record “We allow extra options to be specified in a DKIM-Signature header, but do not allow extra options to be specified in a DKIM TXT record. (I don't recall this being discussed before, but just may not remember it.) Should we? If not, how would we do upwardly-compatible changes without requiring multiple DNS entries for both an old and new entry.” [Described as part of tag-list syntax, §3.2: “Unrecognized tags MUST be ignored.”] http://mipassoc.org/pipermail/ietf-dkim/2006q1/002260.html issue with DKIM simple header algorithm and milter-based implementations 1204 tony(_at_)att(_dot_)com OPEN seemed like consensus but no clear change Q about milter handling of white space around colons in headers [I have a sendmail patch to fix this] http://mipassoc.org/pipermail/ietf/dkim/2006q1/002273.html clarifications on use of l= tag 1215 Eric Allman OPEN no discussion http://mipassoc.org/pipermail/ietf/dkim/2006q1/002185.html (bad URL) (item was confirmation of language inserted into draft) signature h= and z= tags 1216 Hector Santos OPEN little discussion Can the lists differ? [probably SHOULD NOT] If they do, which one wins? [h=] Why so complex? http://mipassoc.org/pipermail/ietf-dkim/2006q1/002375.html ABNF: Sender = Originator / Operator 1222 dhc(_at_)crocker(_dot_)net OPEN (also listed as 1221) some discussion Summary: never use the word “sender” ever again (use “originator” or “operator” instead) http://mipassoc.org/pipermail/ietf-dkim/2006q1/002495.html DKIM and mailing lists 1224 Stephen Farrell OPEN too much discussion http://mipassoc.org/pipermail/ietf-dkim/2006q1/002534.html http://www.sympa.org/wiki/doku.php?id=dkim_and_mailing_lists http://mipassoc.org/pipermail/ietf-dkim/2006q1/001839.html 512 too short? 1226 Stephen Farrell OPEN some discussion Summary: RSA key size should be 1024 minimum http://mipassoc.org/pipermail/ietf-dkim/2006q1/002620.html bunch of nits for base 1227 Stephen Farrell OPEN no discussion http://mipassoc.org/pipermail/ietf-dkim/2006q1/002615.html Why is s= REQUIRED? 1228 Stephen Farrell OPEN a tiny bit of discussion Summary: shouldn’t there be a default selector? http://mipassoc.org/pipermail/ietf-dkim/2006q1/002621.html z= field and EAI wg 1229 Stephen Farrell OPEN a tiny bit of discussion “Even if it doesn't hit anywhere else, presumably the EAI work will have to be taken into account for the z= field, with potential changes being required to the current ABNF?” http://mipassoc.org/pipermail/ietf-dkim/2006q1/002622.html selectors and key rollover 1230 Stephen Farrell OPEN no discussion Summary: Version numbers on selector names Multiple keys per selector http://mipassoc.org/pipermail/ietf-dkim/2006q1/002619.html some process-problematic references in base 1231 Stephen Farrell OPEN no discussion Summary: Search for DKK first creates problematic reference (skip this and revise doc later?) Authentication-Results [should already be gone] §6.6 (MUA Considerations) — necessary/useful? http://mipassoc.org/pipermail/ietf-dkim/2006q1/002616.html Clarify delegation to 3rd parties N001 Stephen Farrell OPEN no discussion “I'd like there to be a very clear consensus as to what's included here, e.g. we are not going to mandate who generates keys, so we thus cannot say whether a private key is being used for >1 sending domain. As it is, the feature is mentioned a number of times, without ever really saying what's to be supported. “That may create potential holes. The problem is that there might be many of those. Is there any way that this feature could be separated out into some kind of extension spec? Anyway, perhaps a section specific to delegation should be added?” http://mipassoc.org/pipermail/ietf-dkim/2006q1/002618.html base editorial N002 Stephen Farrell OPEN no discussion Move “some of the text here” [?] to overview document Provide examples at the beginning of the document to make it easier to understand http://mipassoc.org/pipermail/ietf-dkim/2006q1/002617.html Analyzing Failures: List of Possible Reasons N003 Hector Santos OPEN “I think section 6.5 is a good step but we need a section that is dedicated to all the possible reasons for failures as we KNOW it to possibly to occur. I think there should a special section: 6.6 List of Possible Failures …” http://mipassoc.org/pipermail/ietf-dkim/2006q1/002694.html