Internet Engineering Task Force 100, 15 Nov 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF100-DCRUP-20171115-0930

Description

DCRUP meeting session at IETF100
2017/11/15 0930

https://datatracker.ietf.org/meeting/100/proceedings/

A

B

Big hand is on the 11 and the little hand is on the 30 shouldn't.

A

It be between them shouldn't the big hit a little handy between them right.

B

This is the eye I'm told I got that backwards, whatever.

A

This is the D craft working group. If this is the wrong flight for you, the adores back there.

A

No well slides the usual Oates's our agenda for all right, we'll do it that way. This is the note, please note the note: well, your obligations, etcetera under intellectual property and so forth. The ITF our agenda quickly bash. Otherwise we can just get started. John is pretty much gonna run. The show for us here today, I believe the dkm usage document is with the RFC editor now. Thank you. Everybody for helping and contributing the DCAM ECC document.

A

I think we have consensus that we are going to let this drop I got that right, let's see all nods and so forth, so we're the last document, then is John and John is in that document. The last document is dekum. Crypto and john is going to leave the discussion on open issues.

B

I'll point out that that was nine minutes of the agenda done in about nine seconds.

A

Except for that time that you said we're gonna, look at dot the document and it turned it out. The document was in last call already.

C

Well, we can try and do that.

A

C

There are, as far as I know, two quick two remaining questions on this document. The the larger question is, it defines two: it defines two new to new signing algorithms, one is RSA with hash keys and the other is EB to five five, one. Nine and.

C

Yeah hold on a second there. We go how about about that? Okay, yeah, hi, Barry, yes defines two algorithms. There is. There still seems to be some question whether we need both of them since they both solve sort of the same problem, except they solve it differently, assuming that one of the assuming that we are still interested in the the elliptical curve algorithm.

C

There are two versions of the of there's two versions of EB to five: five, one, nine there's one that there's the pure version and there's the hashed version, and my my document has has said to use the pure version, because the the canonicalization already hashes stuff. However, everybody who knows about these crypto libraries tells me that the only one that if anybody is actually the implement, is the hashed version.

C

Okay, and so after after beating back some suggestions like oh, you can simply unwind all of the code that does the hashing and give us the yeah people. People who know more about know more about crypto than I.

C

Do you think tell me that rehashing the hash will make will waste time, but won't make make things any worse, so so the the proposal now is where it says to use the pure version of edy, two five: five, one: nine in fact we're going to use the hashed version and even though the input to it is or is already hashed. So the first question is: everybody: is everybody who's ever? Do you think that is fine and then we're gonna back up and the question is okay or we got?

C

Are you actually gonna go to gonna do both of these, and if the answer is no, which one do you prefer, so why not? Why don't we my dispose of the second one? First, since there seems to be no, everybody seems to agree like fine hash hash it twice whatever.

C

A

Thumbs down dit, but he's not getting up to the mic. No.

C

No Jim didn't do the thumbs up. Oh yeah, I'm sorry I saw.

A

D

C

A

C

The other question is: are we gonna do both? Should we do both of these, or should we only do one? You know and I sent out yeah I sent out a note explaining the advantages and disadvantages of both I mean both either either. One of them solves the problem that that broken DNS provision systems can't handle lark can't handle long key strings in text records.

C

So the bad you see the bad thing about about ECC is that the libraries aren't. The libraries are only only slowly becoming slowly becoming available, so we'll take a lot before we can do it. The bad thing about about hash keys is that they're there there is. There is a patent, the status of which we don't know I gather, because, even though the author is in the room, he no longer works for the company that owns the patent rights.

C

No, it's cisco, you know who historically have been pretty reasonable, but we don't know so that's where we are now I, encourage you or up to me, I would just say: leave them both in it's fine, but since we're all here in the room, I encourage people I encourage anybody who disagrees with me or even who reads with me to to tell me why I am right or wrong.

E

All right all bye, this is Jim Fenton. So, in addition to the IPR issue and I guess, my preference is for is for ECC of the two just because I would just not like to have my name involved with an IPR issue, partly yeah. Okay, your.

C

Name's not in the draft. We just know who you are right.

E

Yeah well actually, my name's, my name's on the the.

C

D

Disclosure right.

E

So I don't like the idea of doing both simply because it it makes potentially makes implementations bigger. It's it's one more reason not to interoperate and I. Would you know if we've got one way of doing something that works? We should pick one and I don't feel strongly about which one, but you know I told you my preference and why so that's that's. Where I stand. Okay,.

F

You know Jenkins uh I I just agree that I think in general, it's better to pick one than to have two options. If we think they're both fine, then it doesn't matter which one we pick. But again it reduces the amount of implementation overhead and also that increases the chance to operability yeah.

C

It does while you're standing there, if you had to pick one which one would you pick I.

F

Haven't read them enough detail to pick one yet sorry out, Jim.

C

I hope actually room. There's enough attention we're talking about the two easy algorithms. No, no! No we're talking about our Thea RSA versus IDI.

G

Do IDI RSA's, okay,.

F

Neal again yeah not having read them in enough detail it's bit, but I would general probably agree that I would pick ECC over our say in hashing yeah.

C

I think it's fair to say that the implementations are with it with and within the D chemical. The implementations are, you know within three lines, at the same complexity as calling different libraries I mean. So it's it's entirely. A crypto issue.

A

Does anyone eat one or art Barry? This.

B

Is Barry, so I suggest that the chairs post a message to the list asking if there are any objections to just going with ECC and see where we get? That would be fine with me. I.

C

Will be you know if that's what we decide? I can certainly chop out chop out the dead language in a matter of minutes, so.

D

Yeah consensus on the room to be confirmed on the list to elliptic curve yeah.

C

Yeah- and they looked at curve, we will do- is the double hashed version for implementation reasons right.

D

Don't think of it as a double hash. Think of the signature as a black box yeah the fact that the signature does a hash yeah.

G

D

You're presenting to be signed is a half.

C

Is a digest, except that we are then using the non pure version that you know I think I can get the language right now. Yeah.

B

Right and so for the mailing list, I think rich can get the language right for the question on the mailing list and and I specifically suggest that we just ask for objections. Okay, not go any further. Okay,.

E

And I said specifically like the way that rich described it just now, you know signed a digest, I think that's that's very clear and and doesn't make it seem like we're doing something kind of weird something that should raise the the concerns of the crypto people. Okay,.

C

Okay with me, anything else.

C

um Well, my direction seems to be to wait for you guys to send the message and unless, unless I hear screaming from the from North America chop out the dead bits and we're done.

A

Does anybody recall ecers preference, he lamented I'm sure limit? That is a strong word couldn't be here today because of the conflict with another working group, but obviously wants to know what we decide here. So eka.

C

Spent a lot of time insisting that if we were going to, if we were going to do key hashes, we needed to do key hashes everywhere. Okay, but now we're gonna do key hashes, nowhere so I. My recollection is that he's fine with Edie yeah and.

B

I, don't know I, guess: Phil can't use the Medeco cue, so he's posted something in jabber and he says once you are signing a digest, you lose the alleged proof on the direct signature approach anyway,.

C

Like yeah I would Martin Thompson said that there I think this was a malicious version of the draft, but I'm not sure I posted it. There is an obscure attack where you can reverse engineer a key from the signature, and even if it's not the right key, it's a key that will work, which is why he thinks, which is why he looked. He liked the thing that put the key in the signature, so you could make sure that you hadn't done this reverse engineering. Rsa is subjective. This I don't know where the IDI is.

D

No, the deterministic DSA isn't if it's the one I'm thinking of but okay.

A

D

So anyone have or imagine any new business that the working group should take up.

A

Okay, I'm in the middle of marking the crypto document. It's dead.

D

B

C

This is not visit and anybody who anybody who knows people over down down the virtual Hall in the Demark group? Sorry they might we if we're gonna, add these to D came. We should probably add him to arc I.

B

Would suggest that John go talk to Kurt and Seth who are working on the arc document and put that bug in their ear and perhaps work with them on wording. I talk.

C

Does that all the time that should be easy.

A

Since we're nearing the end of processing, the crypto document will be I'll, be looking for a shepherd. If there's anybody who's dying, they do the Faubert hold it. Okay, thank you all, but you're on there.

D

Okay, then, unless we anticipate any new work, this is thanks thanks for coming to the second and final meeting, the working group. Oh no.

G

G

I'm very did we just start earlier this year? Yes, yeah I think that's very fine to have working groups like that I assume I'm going to get the publication requests if things go well before Christmas, okay, no just did I'm just trying to do my own planning of you know when I'll go get that all right. Thank you.

H

D

You all yeah, if you haven't right here, that's only the one yeah, so you haven't signed blue sheet, come up to the front of the room, otherwise go to lunch.