[{"author": "Michael Hammer", "text": "<p>As we get close to starting could someone share how many people are physically in the room? Can't tell from the camera.</p>", "time": "2025-07-22T15:00:11Z"}, {"author": "Murray Kucherawy", "text": "<p>About 20.</p>", "time": "2025-07-22T15:00:20Z"}, {"author": "Michael Hammer", "text": "<p>Thanks, Murray.</p>", "time": "2025-07-22T15:00:34Z"}, {"author": "Al Iverson", "text": "<p>Audio sounds good.</p>", "time": "2025-07-22T15:01:37Z"}, {"author": "Dave Crocker", "text": "<p>Blue dots by a participant name seems to indicate they are a chair.  Gosh but there are a lot of chairs.</p>", "time": "2025-07-22T15:03:53Z"}, {"author": "Murray Kucherawy", "text": "<p>Indeed.  Good to see the bench has grown.</p>", "time": "2025-07-22T15:04:15Z"}, {"author": "Dave Crocker", "text": "<p>In some organizations, the accepted fig leaf for reaming an action 'constructively' is to start with the phrase \"Even better if...\"</p>", "time": "2025-07-22T15:09:26Z"}, {"author": "Bron Gondwana", "text": "<p>Yes and</p>", "time": "2025-07-22T15:09:46Z"}, {"author": "Bron Gondwana", "text": "<p>I don't like half of your documents half as well as you deserve</p>", "time": "2025-07-22T15:10:58Z"}, {"author": "Trent Adams", "text": "<p>I like the \"no opinion\" option... :)</p>", "time": "2025-07-22T15:11:19Z"}, {"author": "Murray Kucherawy", "text": "<p>The documents that were supposed to sack the candidate documents have been sacked.</p>", "time": "2025-07-22T15:11:37Z"}, {"author": "Dave Crocker", "text": "<p>@trent, 'no opinion' is almost a hostile choice, in an IETF context.</p>", "time": "2025-07-22T15:11:53Z"}, {"author": "Trent Adams", "text": "<p>Llamas!</p>", "time": "2025-07-22T15:11:54Z"}, {"author": "Pete Resnick", "text": "<p>I'm inclined to let Richard keep going until someone else gets in the queue? Anyone else have comments?</p>", "time": "2025-07-22T15:13:54Z"}, {"author": "Michael Hammer", "text": "<p>Works for me.</p>", "time": "2025-07-22T15:14:17Z"}, {"author": "Pete Resnick", "text": "<p>Thanks Dave. :-)</p>", "time": "2025-07-22T15:14:30Z"}, {"author": "Dave Crocker", "text": "<p>mumble.</p>", "time": "2025-07-22T15:14:37Z"}, {"author": "Murray Kucherawy", "text": "<p>Vague idea: If Bron's document addresses N problems, should we consider addressing them in N distinct shorter but related drafts rather than what amounts to an almost-omnibus draft?</p>", "time": "2025-07-22T15:15:38Z"}, {"author": "Michael Hammer", "text": "<p>A very UNIX like approach.</p>", "time": "2025-07-22T15:16:17Z"}, {"author": "Bron Gondwana", "text": "<p>I will speak to that when the queue gets to me :)</p>", "time": "2025-07-22T15:16:36Z"}, {"author": "Murray Kucherawy", "text": "<p>I imagine the risk is successfully doing some things incrementally but eventually still reaching a point where that's no longer possible to address everything completely.</p>", "time": "2025-07-22T15:17:29Z"}, {"author": "Trent Adams", "text": "<p>@murray +1 ( I feel multiple shorter drafts are more likely to make it into the light for consideration than churning on a single monolithic draft)</p>", "time": "2025-07-22T15:17:34Z"}, {"author": "Trent Adams", "text": "<p>Can someone send around (or otherwise point out) the email Dave's talking about?</p>", "time": "2025-07-22T15:18:26Z"}, {"author": "Bron Gondwana", "text": "<p>The big risk is getting to the point where you cannot distinguish between legitimate email and attacks in cases where you want to be able to</p>", "time": "2025-07-22T15:18:29Z"}, {"author": "Murray Kucherawy", "text": "<p>@Trent: <a href=\"https://mailarchive.ietf.org/arch/msg/ietf-dkim/-FAXWp73WA-tB_vu7N4WlQKh4NI/\">https://mailarchive.ietf.org/arch/msg/ietf-dkim/-FAXWp73WA-tB_vu7N4WlQKh4NI/</a></p>", "time": "2025-07-22T15:19:11Z"}, {"author": "Dave Crocker", "text": "<p>@murray, there is well-established risk at unecessary fate-sharing, especially when it involves lots of additional complexity.</p>", "time": "2025-07-22T15:19:29Z"}, {"author": "Bron Gondwana", "text": "<p>Pete: for remote speakers, their name also  appears in green at the top of hte list</p>", "time": "2025-07-22T15:19:33Z"}, {"author": "Dave Crocker", "text": "<p>drat. sorry.  what feature did he say dkor is missing?</p>", "time": "2025-07-22T15:21:00Z"}, {"author": "Bron Gondwana", "text": "<p>diff</p>", "time": "2025-07-22T15:21:05Z"}, {"author": "Dave Crocker", "text": "<p>diff actually has nothing obvious to do with the authentication mechanism.  Signing it might be useful, but it is separate function.</p>", "time": "2025-07-22T15:22:15Z"}, {"author": "Bron Gondwana", "text": "<p>I will address that during my voice bit to I guess</p>", "time": "2025-07-22T15:22:41Z"}, {"author": "Murray Kucherawy", "text": "<p>Wouldn't the diff details or some of them might need to become part of the signature?</p>", "time": "2025-07-22T15:22:42Z"}, {"author": "Bron Gondwana", "text": "<p>I disagree with it not being necessary to authentication</p>", "time": "2025-07-22T15:22:51Z"}, {"author": "Bron Gondwana", "text": "<p>@murray they need to be signed; but a signed header is sufficient</p>", "time": "2025-07-22T15:23:21Z"}, {"author": "Bron Gondwana", "text": "<p>it needs to be possible to add NEW diff headers that are unsigned by the previous hops' signatures</p>", "time": "2025-07-22T15:23:44Z"}, {"author": "Murray Kucherawy", "text": "<p>Right.</p>", "time": "2025-07-22T15:23:52Z"}, {"author": "Dave Crocker", "text": "<p>The assertion that there is convergence confuses a small bit of overlap of details with a large bit of non-overlap .</p>", "time": "2025-07-22T15:23:56Z"}, {"author": "Bron Gondwana", "text": "<p>hence the \"header i=N includes diff headers i=1..N\"</p>", "time": "2025-07-22T15:24:09Z"}, {"author": "Dave Crocker", "text": "<p>Changing an installed base at Internet scale (and diversity) is not the same as changing an enterprise, even a very large one.</p>", "time": "2025-07-22T15:25:19Z"}, {"author": "Dave Crocker", "text": "<p>@bron, please explain what the technical basis for linking diff and authentication component function definitions is.</p>", "time": "2025-07-22T15:26:28Z"}, {"author": "Pete Resnick", "text": "<p>I'll be sure to remind Bron to answer that if he doesn't.</p>", "time": "2025-07-22T15:26:57Z"}, {"author": "Michael Hammer", "text": "<p>I think we need to be very careful when we use phrases like \"bad actor\" when talking about a technical standard.</p>", "time": "2025-07-22T15:27:34Z"}, {"author": "Murray Kucherawy", "text": "<p>@Mchael: Want to say why?  at the mic or here</p>", "time": "2025-07-22T15:27:53Z"}, {"author": "Jim Fenton", "text": "<p>Any technical spec that deals with security-related issues needs to consider bad actors.</p>", "time": "2025-07-22T15:28:32Z"}, {"author": "Michael Hammer", "text": "<p>I guess I'll put it this way... \"What is the standard technical definition (normative) of bad actor?\"</p>", "time": "2025-07-22T15:29:10Z"}, {"author": "Jim Fenton", "text": "<p>@wei Please check the minutes wrt what you said on compatibility. I have a sense you may have been misquoted.</p>", "time": "2025-07-22T15:29:36Z"}, {"author": "Murray Kucherawy", "text": "<p>@Michael: Well, you can't really define \"spam\" either.  I liked Gmail's old definition: \"Spam is what our users say it is.\"</p>", "time": "2025-07-22T15:30:33Z"}, {"author": "Michael Hammer", "text": "<p>When we look at DKIM, either a message is properly signed and valid, broken or non-existant.</p>", "time": "2025-07-22T15:30:57Z"}, {"author": "Pete Resnick", "text": "<p>Minutes are here: <a href=\"https://notes.ietf.org/notes-ietf-123-dkim\">https://notes.ietf.org/notes-ietf-123-dkim</a></p>", "time": "2025-07-22T15:31:11Z"}, {"author": "Murray Kucherawy", "text": "<p>I don't think DKIM (or any auth mechanism) can define bad actor, but it can say the algorithm ended in one of N states.  There's no value judgement there.  That's part of local secret sauce.  We can, however, say that we observe bad actors \"using\" DKIM in certain wais.</p>", "time": "2025-07-22T15:31:55Z"}, {"author": "Murray Kucherawy", "text": "<p>ways*</p>", "time": "2025-07-22T15:31:57Z"}, {"author": "Michael Hammer", "text": "<p>I'm not saying we shouldn't use the phrase, simply that we need to be careful.</p>", "time": "2025-07-22T15:32:04Z"}, {"author": "Bron Gondwana", "text": "<p>Not \"toss out\" - \"be able to run in parallel\"</p>", "time": "2025-07-22T15:32:57Z"}, {"author": "Michael Hammer", "text": "<p>@Murray, fair enough. I don't want to go too far down this path.</p>", "time": "2025-07-22T15:33:01Z"}, {"author": "Jim Fenton", "text": "<p>@Mike See RFC 4686 Section 2 for a discussion of bad actors wrt DKIM.</p>", "time": "2025-07-22T15:33:08Z"}, {"author": "Bron Gondwana", "text": "<p>it still uses the keys, which is the major configuration that 99% of people need</p>", "time": "2025-07-22T15:33:12Z"}, {"author": "Dave Crocker", "text": "<p>It might help for Bron an me to have some direct discussion here.  Briefly, of course.</p>", "time": "2025-07-22T15:34:15Z"}, {"author": "Pete Resnick", "text": "<p>Indeed, \"compatibility with installed base\" sometimes means \"works in parallel without breaking old uses\" (which I believe Bron's document does).</p>", "time": "2025-07-22T15:34:22Z"}, {"author": "Pete Resnick", "text": "<p>Dave, if you want to jump in on this specific topic with Bron, I'm fine with that.</p>", "time": "2025-07-22T15:35:13Z"}, {"author": "Dave Crocker", "text": "<p>@pete, you have defined incompatibility.  Tolerating parallel activity has nothing to do with what compatibility means.</p>", "time": "2025-07-22T15:35:19Z"}, {"author": "Michael Hammer", "text": "<p>@Bron, that is a much narrower and clearer approach to \"what is a bad actor\". Thank you.</p>", "time": "2025-07-22T15:36:07Z"}, {"author": "Murray Kucherawy", "text": "<p>Interesting; \"lying into the protocol\" is something we've not had to deal with before.</p>", "time": "2025-07-22T15:36:34Z"}, {"author": "John Levine", "text": "<p>Isn't that what doomed ARC?</p>", "time": "2025-07-22T15:36:52Z"}, {"author": "Pete Resnick", "text": "<p>I would ask for others that agree with Dave's view to chime in to give another take on it. I think that would help all of us.</p>", "time": "2025-07-22T15:37:35Z"}, {"author": "Pete Resnick", "text": "<p>s/that/who</p>", "time": "2025-07-22T15:37:43Z"}, {"author": "Jim Fenton", "text": "<p>Dave's view on which?</p>", "time": "2025-07-22T15:38:03Z"}, {"author": "Pete Resnick", "text": "<p>@Jim: I <em>think</em> (and Dave can correct me if I've mischaracterized) it's that we should have separate documents for each component and that we should develop them independently.</p>", "time": "2025-07-22T15:41:08Z"}, {"author": "Michael Hammer", "text": "<p>From a DKIM perspective it is invalid. That's one of my concerns with calling this DKIM2. We saw similar issues with SPF and SPF2 (SenderID).</p>", "time": "2025-07-22T15:42:16Z"}, {"author": "Dave Crocker", "text": "<p>@murray, I mised the 'lying into' reference.  What was it?\\</p>", "time": "2025-07-22T15:42:56Z"}, {"author": "John Levine", "text": "<p><span class=\"user-mention\" data-user-id=\"3206\">@Michael Hammer</span> I'm OK giving it a different name, but we do plan to use the same DNS key records</p>", "time": "2025-07-22T15:43:08Z"}, {"author": "Dave Crocker", "text": "<p>Richard is engaging in an appeal to authority.</p>", "time": "2025-07-22T15:43:13Z"}, {"author": "Murray Kucherawy", "text": "<p>I feel like for N distinct piecemeal drafts, some subset of N can be done purely incrementally over deployed DKIM, but some other subset can't, and part of the question is whether that's acceptable.</p>", "time": "2025-07-22T15:43:42Z"}, {"author": "Pete Resnick", "text": "<p>Not just authority. I'm not convinced that \"long discussion\" or \"big providers\" is relevant.</p>", "time": "2025-07-22T15:43:43Z"}, {"author": "Pete Resnick", "text": "<p>I'll say so at the mic in a moment.</p>", "time": "2025-07-22T15:44:04Z"}, {"author": "Alexey Melnikov", "text": "<p>Is developing parts independently an editorial thing?</p>", "time": "2025-07-22T15:44:24Z"}, {"author": "Murray Kucherawy", "text": "<p>@Dave: Bron's definition of \"bad actor\" as I heard it involved an actor in the new ecosystem knowingly sending false information into the chain of signatures.  That's what I meant by that.</p>", "time": "2025-07-22T15:44:40Z"}, {"author": "Murray Kucherawy", "text": "<p>@Alexey: I think of it as organizational.</p>", "time": "2025-07-22T15:45:05Z"}, {"author": "Bron Gondwana", "text": "<p>yes, we certainly currently have actors knowingly sending false information into the existing ecosystem</p>", "time": "2025-07-22T15:45:06Z"}, {"author": "Alexey Melnikov", "text": "<p>We can still have multiple docs and require the solution to implement all of them to comply with DKIM2</p>", "time": "2025-07-22T15:45:14Z"}, {"author": "Dave Crocker", "text": "<p>@Pete: Not a separate doc for /each/ of the goals, but separate documents, as makes sense.  For example, note in the Venue column, quite a few references to BCP.  I didn't indicate, but it would not surprise me to find a single BCP sufficient.  This goes to the 'system integration' function I cited, as distinct from the component defintions.</p>", "time": "2025-07-22T15:45:17Z"}, {"author": "Michael Hammer", "text": "<p>@John Levine, I'm just responding to Bron stating that people only validating DKIM vs \"DKIM2\" \"will be getting it wrong\" (I'm paraphrasing).</p>", "time": "2025-07-22T15:46:02Z"}, {"author": "Trent Adams", "text": "<p>IMO - the final solution could be approached via the development of discreet drafts (perhaps with alternate approaches per module)... but... that the full set to be considered must have at least one full path (without gaps).  So, in that vein, I think @Dave's module approave is fine... but I like @Bron's overall approach as the driver for the overall solution.</p>", "time": "2025-07-22T15:46:05Z"}, {"author": "Michael Hammer", "text": "<p>@Murray, agree with your interpretation.</p>", "time": "2025-07-22T15:46:44Z"}, {"author": "John Levine", "text": "<p>@Michael I believe his point is that's why it needs to be a different header. An old DKIM verifier won't use necessary bits to validate properly.</p>", "time": "2025-07-22T15:47:49Z"}, {"author": "Pete Resnick", "text": "<p>Dave: ack. Thanks. There are things in your table that you assign to BCP that seem (to me) belong as protocol, but I at least get the principle.</p>", "time": "2025-07-22T15:48:03Z"}, {"author": "Bron Gondwana", "text": "<p>That's my point yes, a naive DKIM validator will return a negative because it doesn't understand the new fields</p>", "time": "2025-07-22T15:48:23Z"}, {"author": "Dave Crocker", "text": "<p>\"Updating software\" at Internet scale, across all those administrations and implementations, is straightforward in effort and time?:??</p>", "time": "2025-07-22T15:48:45Z"}, {"author": "Murray Kucherawy", "text": "<p>@Bron: But that's true even if old-DKIM or new-DKIM are operating in parallel, right?  old-DKIM signatures will fail where new ones might succeed.</p>", "time": "2025-07-22T15:49:20Z"}, {"author": "Bron Gondwana", "text": "<p>@Murray yes and if we were only using DKIM signature for i=1, that's fine</p>", "time": "2025-07-22T15:51:30Z"}, {"author": "Michael Hammer", "text": "<p>@Murray, exactly. It's not that the implemtors of DKIM validating are naive. It's that for whatever reason they aren't validating for DKIM2.</p>", "time": "2025-07-22T15:51:38Z"}, {"author": "Bron Gondwana", "text": "<p>it's more that we'd have MORE failing DKIM signatures</p>", "time": "2025-07-22T15:51:39Z"}, {"author": "Bron Gondwana", "text": "<p>right now we have \"fails DKIM, passes ARC\"</p>", "time": "2025-07-22T15:52:58Z"}, {"author": "Bron Gondwana", "text": "<p>but for sure you could have things that pass DKIM but fail DKIM2 (because they have been replayed)</p>", "time": "2025-07-22T15:53:21Z"}, {"author": "Dave Crocker", "text": "<p>@bron, since that is exactly why ARC was defined, what is your point?</p>", "time": "2025-07-22T15:53:29Z"}, {"author": "Michael Hammer", "text": "<p>How many receivers are validating for ARC and acting on the results?</p>", "time": "2025-07-22T15:53:55Z"}, {"author": "Bron Gondwana", "text": "<p>@Michael: we can get Richard to speak to that</p>", "time": "2025-07-22T15:54:08Z"}, {"author": "Bron Gondwana", "text": "<p>he describe an example of ARC fraud to me recently</p>", "time": "2025-07-22T15:54:24Z"}, {"author": "Bron Gondwana", "text": "<p>I described how ARC fraud could work 7 years ago after I first read ARC</p>", "time": "2025-07-22T15:55:11Z"}, {"author": "Bron Gondwana", "text": "<p>and it's what the fraudsters are indeed doing</p>", "time": "2025-07-22T15:55:23Z"}, {"author": "Trent Adams", "text": "<p>+1 to @JohnLevine</p>", "time": "2025-07-22T15:56:12Z"}, {"author": "Murray Kucherawy", "text": "<p>Might be helpful to do a draft that \"ends\" the ARC experiment, since it was published as Experimental, like we did with SPF/Sender-ID.</p>", "time": "2025-07-22T15:56:13Z"}, {"author": "Bron Gondwana", "text": "<p>I would be very happy to do an end ARC</p>", "time": "2025-07-22T15:56:28Z"}, {"author": "Michael Hammer", "text": "<p>I get that ARC fraud is out there but if ARC is not generally used, I wonder how relevant and meaningful.</p>", "time": "2025-07-22T15:56:40Z"}, {"author": "Murray Kucherawy", "text": "<p>@Michael: Spending a bit of time to kill it formally might remove a recurring distraction from this area of work.</p>", "time": "2025-07-22T15:57:09Z"}, {"author": "Pete Resnick", "text": "<p>Transition is sounding to me an important issue.</p>", "time": "2025-07-22T15:57:34Z"}, {"author": "Murray Kucherawy", "text": "<p>I think we should start on the AS as soon as we start to get some open experience with implementation.</p>", "time": "2025-07-22T15:58:10Z"}, {"author": "John Levine", "text": "<p>We definitely need some test implementions</p>", "time": "2025-07-22T15:59:03Z"}, {"author": "Jim Fenton", "text": "<p>Yes. If DKIM2 is a separate header field, there needs to be some guidance on also requiring a DKIM signature as well until some threshold of DKIM2 adoption is achieved.</p>", "time": "2025-07-22T15:59:13Z"}, {"author": "Michael Hammer", "text": "<p>I think \"transition\" is an important issue, Pete. I also agree with what Dave just said.</p>", "time": "2025-07-22T15:59:19Z"}, {"author": "Dave Crocker", "text": "<p>sorry.  who is speaking?</p>", "time": "2025-07-22T15:59:30Z"}, {"author": "Phillip Tao", "text": "<p>Re: the need for a new header, doesn't the existing DKIM-signature currently require a v= tag? Could we not just rev that version? Senders could insert a v=1 and a v=2?</p>", "time": "2025-07-22T15:59:32Z"}, {"author": "Jim Fenton", "text": "<p>THat was Arnt</p>", "time": "2025-07-22T15:59:46Z"}, {"author": "Dave Crocker", "text": "<p>tnx</p>", "time": "2025-07-22T15:59:56Z"}, {"author": "Pete Resnick", "text": "<p>I think we might talk about the A/S in the context of the \"motivations\" document; turning motivations into the A/S might be possible.</p>", "time": "2025-07-22T16:00:29Z"}, {"author": "Murray Kucherawy", "text": "<p>@Phillip: Use of \"v=\" for this has been contentious.</p>", "time": "2025-07-22T16:00:34Z"}, {"author": "Dave Crocker", "text": "<p>The choice is not between 'doing it' vs. 'not doing it'.  It is between trying to do it gracefully versus traumatically.</p>", "time": "2025-07-22T16:00:38Z"}, {"author": "Jim Fenton", "text": "<p>@phillip the problem is that the semantics of the v= field aren't well defined and are probably misimplemented by many.</p>", "time": "2025-07-22T16:01:06Z"}, {"author": "Michael Hammer", "text": "<p>@Dave, +1</p>", "time": "2025-07-22T16:01:10Z"}, {"author": "Murray Kucherawy", "text": "<p>The parsing/generation and DNS logic should be easy to separate out from DKIM, even if under a different name.</p>", "time": "2025-07-22T16:01:48Z"}, {"author": "Michael Hammer", "text": "<p>+1, Murray</p>", "time": "2025-07-22T16:02:11Z"}, {"author": "Alexey Melnikov", "text": "<p>Right, we can reuse DKIM signing engine for new work</p>", "time": "2025-07-22T16:02:37Z"}, {"author": "Phillip Tao", "text": "<p>Well, hopefully then, we can define a more meaningful version tag this time around</p>", "time": "2025-07-22T16:02:40Z"}, {"author": "John Levine", "text": "<p>The DNS logic is inherited from domainkeys</p>", "time": "2025-07-22T16:02:40Z"}, {"author": "Murray Kucherawy", "text": "<p>@Phillip: The more general claim is that version tags aren't a good idea.</p>", "time": "2025-07-22T16:03:02Z"}, {"author": "Murray Kucherawy", "text": "<p>As I recall it goes something like: If you've changed the protocol enough to need a new \"v=\" value, then it's really a new protocol altogether, and you should call it something different.</p>", "time": "2025-07-22T16:03:40Z"}, {"author": "Bron Gondwana", "text": "<p>Version tags are the same as new header fields IFF you can't validate a higher version key with an old verifier</p>", "time": "2025-07-22T16:03:48Z"}, {"author": "Murray Kucherawy", "text": "<p>Right, that</p>", "time": "2025-07-22T16:03:59Z"}, {"author": "Murray Kucherawy", "text": "<p>On the other hand, if the new thing is fully compatible with the old thing, then you don't need a version tag.</p>", "time": "2025-07-22T16:04:15Z"}, {"author": "Bron Gondwana", "text": "<p>so you have to do both versions anyway, and if you're doing that a new header field name is clearer</p>", "time": "2025-07-22T16:04:15Z"}, {"author": "John Levine", "text": "<p>Version tags turn out not to work, as soon as you have a new version in practice you have a new scheme and you might as well give it a new name. In this case we will probably be putting both DKIM2 and DKIM signatures on mail for quite a while</p>", "time": "2025-07-22T16:04:25Z"}, {"author": "Dave Crocker", "text": "<p>Number of independent software implementations is only one of the sources of complexity (or simplification.)  Rates of updating vary enormously.  And then there is the range of configuration options by each platform using an implementation.</p>\n<p>Combinatorials at Internet scale are never simple.</p>", "time": "2025-07-22T16:04:46Z"}, {"author": "Michael Hammer", "text": "<p>@John Levine, we saw that with DK and DKIM.</p>", "time": "2025-07-22T16:05:28Z"}, {"author": "Dave Crocker", "text": "<p>@murray, @bron, re: version tags.  Yup.</p>", "time": "2025-07-22T16:05:41Z"}, {"author": "Phillip Tao", "text": "<p>Consider me convinced</p>", "time": "2025-07-22T16:06:39Z"}, {"author": "Murray Kucherawy", "text": "<p>I do sympathize with the apparent inelegance of needing a new header field name every time though.  I hope we don't end up registering \"DKIMn-Signature\" as a header field name for n in some range.</p>", "time": "2025-07-22T16:07:30Z"}, {"author": "Michael Hammer", "text": "<p>ARC was an interesting experiment.</p>", "time": "2025-07-22T16:07:53Z"}, {"author": "Bron Gondwana", "text": "<p>if we don't get it right within by version 6, for sure we can consider the name cursd</p>", "time": "2025-07-22T16:07:55Z"}, {"author": "Bron Gondwana", "text": "<p>As pointed out with IP, nothing with version 6 gets adoption</p>", "time": "2025-07-22T16:08:16Z"}, {"author": "Dave Crocker", "text": "<p>The time it takes to write the new code is pretty close to irrelevant, in the scheme of Internet-scale adoption.</p>", "time": "2025-07-22T16:08:25Z"}, {"author": "Pete Resnick", "text": "<p>heh</p>", "time": "2025-07-22T16:08:25Z"}, {"author": "Murray Kucherawy", "text": "<p>@Bron: DKIMSEC</p>", "time": "2025-07-22T16:08:29Z"}, {"author": "Bron Gondwana", "text": "<p>I do appreciate the \"second system\" comments about this though</p>", "time": "2025-07-22T16:08:33Z"}, {"author": "Bron Gondwana", "text": "<p>@Murray - just because Pete isn't on the ombudsteam any more doesn't mean you can just say things like that.</p>", "time": "2025-07-22T16:08:59Z"}, {"author": "Murray Kucherawy", "text": "<p>@Bron: Might be worth a paragraph in at least one of the documents just to say \"Yes, we thought of this.\"</p>", "time": "2025-07-22T16:09:03Z"}, {"author": "Pete Resnick", "text": "<p>Is *SEC now a dirty word?</p>", "time": "2025-07-22T16:12:14Z"}, {"author": "Dave Crocker", "text": "<p>@Pete, no.  It has always been a dirty word.</p>", "time": "2025-07-22T16:13:04Z"}, {"author": "Pete Resnick", "text": "<p>:-D</p>", "time": "2025-07-22T16:13:15Z"}, {"author": "Dave Crocker", "text": "<p>@pete, 50 years ago, sec folk said that sec had to be difficult to use.  Had to be.</p>", "time": "2025-07-22T16:14:19Z"}, {"author": "Kenneth Murchison", "text": "<p>NOT it for editing an A/S</p>", "time": "2025-07-22T16:14:42Z"}, {"author": "Bron Gondwana", "text": "<p>Have you checked with your Boss's Boss on that Ken?</p>", "time": "2025-07-22T16:15:00Z"}, {"author": "Murray Kucherawy", "text": "<p>I feel some PTSD in Ken's recent comment.</p>", "time": "2025-07-22T16:15:55Z"}, {"author": "Michael Hammer", "text": "<p>Ha</p>", "time": "2025-07-22T16:16:15Z"}, {"author": "Dave Crocker", "text": "<p>pubkey in the message with hash in the DNS is authenticating the key for authentication.  So its keys all the way down.</p>", "time": "2025-07-22T16:19:53Z"}, {"author": "Jim Fenton", "text": "<p>Hash in the DNS is really an authorization for the key, not authenticaiton.</p>", "time": "2025-07-22T16:20:31Z"}, {"author": "Dave Crocker", "text": "<p>@jim, sure, but that's not a humorous.</p>", "time": "2025-07-22T16:20:52Z"}, {"author": "Dave Crocker", "text": "<p>a -&gt; as</p>", "time": "2025-07-22T16:21:00Z"}, {"author": "Bron Gondwana", "text": "<p>There was discussion of doing merkle trees all the way down in DISPATCH the other day</p>", "time": "2025-07-22T16:21:01Z"}, {"author": "Bron Gondwana", "text": "<p>so you'd just put a pointer to the key in the certificate transparency chain rather than the public key itself</p>", "time": "2025-07-22T16:21:27Z"}, {"author": "Murray Kucherawy", "text": "<p>I think we had text in 6376 or similar that acknowledged that email people and DNS people generally don't like to interact very much.</p>", "time": "2025-07-22T16:21:40Z"}, {"author": "Jim Fenton", "text": "<p>@dave Sorry, humor doesn't come through in text all that well</p>", "time": "2025-07-22T16:21:55Z"}, {"author": "Dave Crocker", "text": "<p>@bron, so it's all the way down, all the way down.</p>", "time": "2025-07-22T16:22:04Z"}, {"author": "Bron Gondwana", "text": "<p>@dave I concur that the A/S is not humorous</p>", "time": "2025-07-22T16:22:25Z"}, {"author": "Michael Hammer", "text": "<p>We know that post-quantum is coming so I think whatever this is called should deal with it from the start.</p>", "time": "2025-07-22T16:23:13Z"}, {"author": "Michael Hammer", "text": "<p>And that becomes a strong driver for adoption.</p>", "time": "2025-07-22T16:25:27Z"}, {"author": "Dave Crocker", "text": "<p>Oversigning: there have been multiple comments that the current DKIM method of protecting against adding 'unauthorized' fields is problematic, and that the new proposal fixes that.</p>\n<p>I'd appreciate seeing more text about that onlist and/or in the motivation doc.</p>", "time": "2025-07-22T16:26:25Z"}, {"author": "Murray Kucherawy", "text": "<p>I don't recall seeing the \"problematic\" claims.  I'll have to go look.</p>", "time": "2025-07-22T16:26:53Z"}, {"author": "Dave Crocker", "text": "<p>@murray, advocates for the new approach have been saying the current approach is poor/problematic and the new one is (much) better.</p>", "time": "2025-07-22T16:27:41Z"}, {"author": "Andrew Newton", "text": "<p>@murray, I have an update on the pq question if you have time</p>", "time": "2025-07-22T16:27:58Z"}, {"author": "Murray Kucherawy", "text": "<p>@Andrew: ack</p>", "time": "2025-07-22T16:28:19Z"}, {"author": "Allen Robinson", "text": "<p>I think it's more just awkward. If it's always the \"right thing\" to oversign every field, having to say that in h= is slightly less good than changing the hashing algorithm to do it automatically</p>", "time": "2025-07-22T16:29:00Z"}, {"author": "Dave Crocker", "text": "<p>@murray, what worries me is that I don't recall ever seeing complaints from the community about the current method.</p>", "time": "2025-07-22T16:29:15Z"}, {"author": "Murray Kucherawy", "text": "<p>We made the current system because we wanted to tolerate header field removal by some MTAs (I recall Exchange being named, but I could be wrong).  Has that aspect of the ecosystem changed?</p>", "time": "2025-07-22T16:29:43Z"}, {"author": "Murray Kucherawy", "text": "<p>@Dave: Same, but Allen may have just summarized it.</p>", "time": "2025-07-22T16:30:11Z"}, {"author": "Dave Crocker", "text": "<p>@allen, awkward, of course, is different from problematic.  One of the characteristics of second system syndrome is trying to fix things that are a matter of aesthetic preference rather than real need, and thereby add complexity that increases risk.</p>", "time": "2025-07-22T16:30:36Z"}, {"author": "Murray Kucherawy", "text": "<p>But then you don't need general oversigning, you just need logic like \"Were there any header fields present that were not part of the hash?\"</p>", "time": "2025-07-22T16:30:50Z"}, {"author": "Dave Crocker", "text": "<p>@murray, my point is not that the new approach isn't cleaner, or whatever, but that it introduces change that is not demonstrably needed.  And change means risk.</p>", "time": "2025-07-22T16:31:40Z"}, {"author": "Jim Fenton", "text": "<p>Not <em>any</em> header fields, just those header fields that were oversigned</p>", "time": "2025-07-22T16:31:45Z"}, {"author": "Michael Hammer", "text": "<p>Apart from being sloppy, most people won't generate a custom list, even if you allow it. Is that a corner case?</p>", "time": "2025-07-22T16:32:09Z"}, {"author": "Dave Crocker", "text": "<p>\"one chance to change dkim\"  = \"one chance to change IP\".  Same outcome.</p>", "time": "2025-07-22T16:32:17Z"}, {"author": "Michael Hammer", "text": "<p>+1, Richard.</p>", "time": "2025-07-22T16:33:53Z"}, {"author": "Jim Fenton", "text": "<p>If we sign most headers, it won't be practical to list all of the header field names unless we want a very large DKIM2 signature header field</p>", "time": "2025-07-22T16:35:18Z"}, {"author": "Dave Crocker", "text": "<p>@jim, while ugly -- which things already were and will remain -- how is the length of the field a problem?</p>", "time": "2025-07-22T16:36:13Z"}, {"author": "Jim Fenton", "text": "<p>It's mostly a problem for geeks like me that look at the signature, I admit.</p>", "time": "2025-07-22T16:36:51Z"}, {"author": "Michael Hammer", "text": "<p>+1, Dave</p>", "time": "2025-07-22T16:40:11Z"}, {"author": "Dave Crocker", "text": "<p>tnx</p>", "time": "2025-07-22T16:40:58Z"}, {"author": "Al Iverson", "text": "<p>+1 to FUSSP as an RFC, if Dave wants to pursue that, sure.</p>", "time": "2025-07-22T16:41:05Z"}, {"author": "Murray Kucherawy", "text": "<p>I'd sponsor it if I could.</p>", "time": "2025-07-22T16:41:26Z"}, {"author": "Allen Robinson", "text": "<p>FWIW I tend to agree, and I think I've said this onlist. My opinion is that changing the semantics of h= from what DKIM specifies isn't required to make DKIM2 work. But if changes are being made anyway, maybe it's worth considering.</p>", "time": "2025-07-22T16:41:32Z"}, {"author": "Bron Gondwana", "text": "<p>@jim the idea for sure is to have the list be unsigned-headers rather than signed-headers; at that point</p>", "time": "2025-07-22T16:41:46Z"}, {"author": "Dave Crocker", "text": "<p>@murray, now to see whether Corey is amenable.</p>", "time": "2025-07-22T16:41:58Z"}, {"author": "Bron Gondwana", "text": "<p>protocol grease</p>", "time": "2025-07-22T16:42:20Z"}, {"author": "Murray Kucherawy", "text": "<p>A bit louder, Mike?</p>", "time": "2025-07-22T16:42:47Z"}, {"author": "Richard Clayton", "text": "<p>I don't think FUSSP is Doctorow ... it's Schryver</p>", "time": "2025-07-22T16:43:13Z"}, {"author": "Murray Kucherawy", "text": "<p>@Andrew: So we can take it up if we manage to drag this out long enough? :)</p>", "time": "2025-07-22T16:43:20Z"}, {"author": "Bron Gondwana", "text": "<p>this is a real question: what's more valuable: compatibility with existing keys, or better keys</p>", "time": "2025-07-22T16:43:32Z"}, {"author": "Dave Crocker", "text": "<p>Doctorow's page on anti-spam proposals:</p>\n<p><a href=\"https://craphound.com/spamsolutions.txt\">https://craphound.com/spamsolutions.txt</a></p>", "time": "2025-07-22T16:45:45Z"}, {"author": "Andrew Newton", "text": "<p>Dragging \"this\" out long enough is not a goal. :)</p>", "time": "2025-07-22T16:45:46Z"}, {"author": "John Levine", "text": "<p>@Richard there are two different ones. Vern's is here <a href=\"https://www.rhyolite.com/anti-spam/you-might-be.html\">https://www.rhyolite.com/anti-spam/you-might-be.html</a></p>", "time": "2025-07-22T16:46:26Z"}, {"author": "Dave Crocker", "text": "<p>@bron, false dichotomy.  Facilitating transition that will take years/decades is a goal.</p>", "time": "2025-07-22T16:46:39Z"}, {"author": "Bron Gondwana", "text": "<p>@Dave: it's more \"where do we start\"?</p>", "time": "2025-07-22T16:46:59Z"}, {"author": "Bron Gondwana", "text": "<p>do we draw a baseline above existing DKIM2 RSA keys or not</p>", "time": "2025-07-22T16:47:11Z"}, {"author": "John Levine", "text": "<p>@richard I think we mean this one <a href=\"https://slashdot.org/comments.pl?sid=98024&amp;cid=8373855\">https://slashdot.org/comments.pl?sid=98024&amp;cid=8373855</a></p>", "time": "2025-07-22T16:47:15Z"}, {"author": "Dave Crocker", "text": "<p>@bron, start where we are, not where we want to be.</p>", "time": "2025-07-22T16:47:18Z"}, {"author": "Murray Kucherawy", "text": "<p>Yeah, I think I agree with Bron: should we prepare for this eventuality, or punt it and retrofit it later</p>", "time": "2025-07-22T16:47:23Z"}, {"author": "Bron Gondwana", "text": "<p>Dave: I think you're agreeing with me on \"allow existing keys to keep working\".</p>", "time": "2025-07-22T16:47:45Z"}, {"author": "Murray Kucherawy", "text": "<p>At a minimum, don't do anything that precludes that being added later.</p>", "time": "2025-07-22T16:47:50Z"}, {"author": "Bron Gondwana", "text": "<p>which means we don't mandate PQ</p>", "time": "2025-07-22T16:47:53Z"}, {"author": "Dave Crocker", "text": "<p>@bron.  Agreeing with you?  Am I allowed to do that?</p>", "time": "2025-07-22T16:48:10Z"}, {"author": "Bron Gondwana", "text": "<p>@Dave I think we have a daily quota and we might be OK</p>", "time": "2025-07-22T16:48:29Z"}, {"author": "Bron Gondwana", "text": "<p>so long as nobody make a quorum call</p>", "time": "2025-07-22T16:48:35Z"}, {"author": "Dave Crocker", "text": "<p>@bron, whew.</p>", "time": "2025-07-22T16:48:41Z"}, {"author": "Bron Gondwana", "text": "<p>I will upload the current motivation doc with the new name, knowing it will probably have a revision soon</p>", "time": "2025-07-22T16:53:07Z"}, {"author": "Murray Kucherawy", "text": "<p>@Bron: Groovy, thanks!</p>", "time": "2025-07-22T16:53:25Z"}]