Internet Engineering Task Force 98, 27 Mar 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF98-CURDLE-20170327-1710

Description

CURDLE meeting session at IETF98
2017/03/27 1710

A

B

C

Yeah I think I'll do all the talking this one floods yeah does.

D

Anybody want to be the jabber scribe. Thank you, yo. Anybody want to take minutes. It's really easy, really, really really easy. I'll give you a cough drop. I'll. Give you to cough drops seriously. We can't proceed DJ, all right, Melinda! Thank you. Anyone who goes to certificate transparency now as to volunteer to take minutes.

A

D

Know there are the stickers this time do they need to lose you yes,.

E

F

It one hour what.

D

Take all my cough drops.

D

It like to be kinda easy if you just download the PDF or something and add annotations to it, because we are going to we're going to try to go really quickly so that everyone can get to the github off, which I think it really doesn't.

D

Alright, the blue sheets are going around there on that side of the room, that's them over by now. You have all seen the note well at least three times. Oh my now yeah, okay, there we go now, I'm using the mic. I am rich. This is daniel 60, so you've all seen to know. Well, basically, it says if you've got any IP r, you must have disclosed that you've got the IPR. Otherwise, please leave the link.

D

Okay, all right Daniel did this bang-up job of coming up with a slide template to discuss all of the drafts so.

D

Without further ado, most of this meeting is probably going to be asking for people to commit to reading the drafts there in a couple cases help Shepherd a document through their short they're, very straightforward, they're, easily separable, there's not a lot of relationships among all of them. It's not like you have to send in understand all of HTTP before you can understand h2. So.

G

Okay, so um this drop I think this dress is one of our priority.

G

We have discussions about assigning your IDs, so we took a decision which is requesting the ini ark and putting this whole idea to under this arc. There has been some discussions about pre hash. This is non pre, ash, signature, I think the consensus is that we should use. We are only defining the draft for non pre hash version variant. So I guess want to comment that I think that that also should be said for all the drive. If we should say that clearly it's it's a basically a design, choice and yeah.

G

There is a awesome identifier for the pre hash variant so well. If we are only focused on the non pre hash, then probably we should remove this next, like I, think there are two slides on this dress.

G

Okay, so the next step, alrighty values, update diversion and we need some reviewers so that this week or next week we can send that to the iesg, so I never launched here.

D

No you care about Sam.

D

Ok, we'll take me to the list. I don't want to do it without at least one up from first. Have you read it I, guess what so it's it's really good shape. Ok,.

G

Tammy yeah.

D

All right girl, CMS.

G

Yes, do you want to go to this one or yeah.

D

This is any discussion.

A

No, no, no, no.

D

You're changing the oids to the I, a narc on republish.

G

I think we will maybe remove everything about the pre hash variant.

G

Other there are the Oh IDs, the algorithm identifier in the beginning, and should we have some comment about the CR else? Also to some of the discussions we had that Sam, the three think I see God.

H

Now you have you a question: have we decided that we're replacing the IEDs that are in the current draft with other royalties they so that we'll have to go into the eyepiece like a need for after about Ed BSA, because it references this one? Okay,.

G

G

Yeah, what's on the story summer, here with you? Okay, so the reason we asked for is that CMS was needed needed to register so mo IDs and we couldn't get the contact information from the current ID used by curdle. So we did not receive answers. We move to an INR. We.

D

Had a short arc donated by symantec in the contact person Oh, otherwise I can find okay.

D

So all right um should we post a draft probably pretty quickly, and then we are ok and then we'll forward it out, though yo yo yo, just sync up with the ipsec box.

A

I

G

um Edss signatures, so nothing really much to say we one person is: we only had one reviewer, so it's pretty shaped but a well. If possible, we would like to have another reviewer. So anyone volunteering.

C

How could somebody devil, for this is open.

D

Yeah I'll get someone okay, so.

G

Next slide, um this one is well has more reviews, I, don't think we need really another reviewer. Any questions regarding this draft.

G

Next draft yeah they're all in Lascaux, so all those um so this draft yeah it's about to use of share too um yeah I. Think we it's it's almost the same. We well. We would need more reviews, especially I, think it's more. The way to graph is written, but it's pretty good. It's pretty well shaped and almost ready, it's a very short drat. I think less than 10 pages. So.

G

People coming later are going to be a sinus reviewers.

I

D

G

Again, if you can't could provide some feedback, Saunders I mean review, choose one or two drafts review them, and then we can look forward.

D

Why don't you get no cookies at the next idea.

A

Oh, come on yeah.

G

Yeah this drafty well, is also pretty much ready. Some changes have been submitted to the least he has been considering various implementations. So I don't think it's very clear, but it would be better to have some other feedbacks. Other reviews, it's mostly to see if the text is clear rather than a contact I mean so that's pretty easy to go. Yeah.

D

It is already widely deployed, discussion bullet, says, and it points up.

G

um Yeah, this one's, so okay, yeah I, get very confused with the name of the draft, so this one is adding a new gfield new key exchange: algorithms based on the new other kernels, the curves yeah. It's only adding a one point or two, so please say you agree with a draft, and so we can move that forward next slide and.

D

If you have any questions on each of these dress or issues you want to give you read it right, and this is you want to bring up step up to the mic and we will back up no problem just trying to set a speed record here. We'll look, the curb don't.

B

Just one question about SSH in general: what happens if you put with the existing applications or app that are out there? If you have nests SH key for an algorithm that isn't understood, did the widely used versions of ssh do something sensible or do they fall over and die? Has anybody looked into it? I.

J

Use a key or a host key, usually, as so I think, like the widely used in Satan plantations, we'll look at a whole bunch of files in your daughter's age directory to try and find something that they can. They can use.

B

D

They're just carry on for the next thing, yeah until it hits the air account. It says too many failures, but yes, it does a reasonable thing.

K

G

Well, this one is also adding two point codes, so.

G

D

On this one too, no.

G

20: okay, it's a good les hommes.

D

This one named adding named eh groups, specific ones, I'm curious, I'm, curious. What TLS folks think about it, since we sort of have kind of dropped, those I.

L

Mean sorry, this is microphone hi and we're kind of like moving towards the different set of groups in this. Some little surprised, you're, not adding the ones that we spent the trouble standardizing. But you know not for us to tell you what to do rather than you should do what we did.

D

So I think this is more about capturing. What's in the deployed base, then moving forward I think that's a valid concern, though, and should probably be reflected in security considerations section this.

F

Is dkg sorry I actually haven't read this at all, but I'm, assuming these are finite field routes that here that's fine I, actually, don't think that you should say use the same groups that we specified for 40 of us, although I do think we should have specific, specified groups. So I'm not sure why ecker recommended that you use the exact same groups that we did. It's fine to have two different protocols use different sets of groups as long as they're strong, primes, usual usual constraints,.

A

J

Of 90 google, so if this is just documenting existing reality than that, that's great and should be done. Oh, why? Wouldn't you just want to use the same groups, just less text.

F

Just to to provide mean the rules for generating groups that are strong are not complicated, like we understand those, and so you might want to not use them across domains. If there's some weird something gets cross domain that we don't understand, there's a bunch of situation with cross domain attacks because you're using a similar protocol and things hook together. It turns out that something TLS looks like something an association can be reused. I, don't see that here, but I don't know, and if you avoid using the same groups, you can avoid that problem.

F

If Greta, these are the same groups as I them, then now you die you're. Just the question is I vs SSH, so I'm totally fine with using one set of groups for one protocol to avoid any possible cross protocol, condemning a condition that I don't understand. I'm not going to have a specific answer. Oh I, don't know I.

J

I didn't find that hugely convincing, but the my organ was purely like it's less typing and that's not great either. Sir.

D

All right we'll try to encourage discussion on the list and maybe pollute I can TLS also.

D

L

Don't think I those groups has any equities here. Like you know, we may have some overall ITF question. What we should do, but it's not like jealous, cares what I mean other than as a general matter of principle cares what I safe does um anything that me adam says these are just this is just you know.

L

This is just endorsing stuff, however, pure to do that great if we're on acceptable, at least if I, if, if we're picking new groups that we nobody use it, we want them to use weepers I ask why, but second I would say like there are these other group structure we that we were happy about.

M

There are q9 sunseeker, and these are the groups that are usually naik and IP. Second, they have been there for all time over and they are venerated in same way that the one certainty Eleazar created, except they are using pi and.

N

The Taylor's.

M

Using me yata care, but ever they're getting ready, I have the same base, so actually either one is fine. It's just you know, which you think is. Both of them are used already, all the donors that we know best right already. So it's just.

D

Picky one so the second half of the phrases and running code. So that would argue for the ike.

D

Okay I, it might be a little premature for last call who solicit some discussion on the list. It might be a bike shed.

D

Anybody else anybody else have any other issues drafts potential work. They want to discuss.

G

Well, we we have um well when the CMS draft and the peak eggs draft will be sent, will we have a receipt of requests to for a GSS update, and so that's going to be the next coming draft.

G

Before we close yeah.

D

That's right so that was off mic there before we close Russ so.

O

I'm sorry I was late. Did we talk about draft p kicks and did we settle on absent null or both, um if not I'd, like the darker added since we're all here, so nope rehash all that I understood? That was.

D

Better than that yeah, no, let's explain the issue. We did not talk about it. We.

O

Did not talk well, it's Jim here, yeah yeah, so the asn.1 structure allowed the.

O

When you're the parameters field in 1997, when they converted from the 1988 syntax to the newer asn.1 syntax, some.

N

Editor dropped.

O

The word optional in algorithm identifier and as a result a bunch of people were forced to specify null as a parameter when none was needed and then they had it optional back in the next release, and so there is some some code that doesn't remember that or does not yet know, hear that optional is there and they insist on some parameter and so I prefer to not do that. But get people update their code and the bottom line is that we have one place in that document.

O

Where there's a discussion on the list about whether it's optional order null and the big point that I have dated benjamin made and I agree with him was pick one don't make it. You know accept either one because then you end up. That's where the interoperability promos come. Am I right Jim.

G

Mirrors did I okay thanks, but I'm. That would be easier if other people can support that view. Yes, underneath.

I

N

I

Warrior I don't.

J

90 google I strongly support that, for you, I don't give a hoot what which one it is. If.

N

J

Doesn't pick one, then we will pick one and we will break everyone else who picks every anything else and and therefore I will pick it so I suggest you pick it.

D

So, on the one hand you'd be like well, the existing deployed base expect something. On the other hand, the existing deployed base is going to need changes to support these curves anyway. So let's just do the clean right thing. Jim. The document right now has picked one.

E

A

Yes, it matters because Java.

E

The standard Java provider can't deal with it a pamphlet and.

E

M

Think they're so document, okay, I, think you're, also a lot of other algorithms that are actually using saying that you must not have you know you first, let's have now they're, so if they don't support that they don't to portos sequence, all girls, either so I think the chiverses need to fix it. Their stuff, fun.

D

I'm not sure how to phrase the question without annoying people, but um then yeah, well, okay, so yeah absent break some Java in both. So how many want to support those hum absence or brick city? Okay, all right! So it's absent simplest, cleanest, break some java, none!

D

No! No! I hated this one s when sequel had the problem too right. No, you carry an explicit I'm, not here and then both so how many people are in favor of supporting both how many people are in favor, having no there's an explicit I'm, not here how many people are in favor of having no indicator when there is no option.

D

That's a pretty clear consensus will confirm it on the list.

A

D

Right, yes, I got to know not and Jim. Do you have to make any doc changes to do that? I.

F

D

Called a straight line: yeah.

H

All right, great.

O

G

D

Right, surprise: okay, anybody else! Anybody else have any other business. They want to bring up.

K

Yes, okay, all.

G

Right there is a last question about the use of PSS and/or PKS s 11.5 so for I. Think it's a RSA.

G

It is sh yeah, so I can find the draft name. Some.

D

Crew, the colonel Arthur, you get shot good one yeah.

G

G

So one question is: do we have a we would like storing to move to PSS, given that well PK CSS is kind of. We can't avoid that one. So do we want to add a PSS, or is it a bad idea to have these two variants, personal.

E

Preference shoot 1.5 in the head.

D

Just like what like TLS 13, did jump.

H

Try you have mirror hitter Goodman on the list said that there's actually no weakness that we know of with PK SI s 1 point 5, so it might as well stay with that. It's true that it's that it's easier to make proofs about PSS, but there's no weakness. Why I add some new stuff when we already have because he s 1 point 5, not my arguments, pretty good yeah.

F

M

H

M

Think they were soon attack, but if you have a very short, you know expert, if you enjoy whispering tree or something like that, there was some type of 1.5 that was more vulnerable to something than PSS yeah. But.

H

I mean I mean if.

M

You are saying that there's nothing for 1.5 there, dearies yeah.

E

M

But I mean I mean yeah, then I'd, basically also actually just moved. A basic was using 1.5 old, also and now in the bin, which defied the Liberator to sleep in search. We actually only departed. You know to say that the PSS system under the soup to speak on which of course causes you know issues because people actually have to implement it when take them into signature algorithm. They can't just make a very simple. You know.

M

Oh, we have the same or sequences up tickle giggle, sickness algorithms, and we just change the format they actually have to implement. Vss also, but.

J

M

Think it actually would be better to move the PSS.

J

Adam Langley Google PSS would be nice, but isn't the world moving away from RSA as a whole, and so is it really worth the time to worry about this.

N

Steaming pile and sorry make up the context wrong, but we're talking about this for Shi yeah, you.

J

P

N

In the room, but then we shouldn't decide.

H

You know I've near jenin opposing you, then, because of TLS 1.2 me epigraphic librarian, the world is going to implement PSS now so might as well use it wherever we have where we have a need for signature, algorithms,.

D

Does anybody have any strong feelings one way or the other.

A

D

Okay, well, we'll take it to the list and the inclinations to say: let's not add something that may not be widely used.

H

G

Thank any questions.

D

All right, yeah, thanks for coming, um we'll, probably wrap up at the next IETF. Everything will just be in the last call next class next idea. For probably that's certainly the last face to face unless somebody comes up with a whole bunch of new drafts and mechanisms. Thank you all for your participation see you later.

D

B

D

Right so with two follow-ups on the list: well, free.

K

P, p, pcs versus already a PSS.

K

What do we do exactly? Because it's.

D

Already on the list, yeah some people, if you haven't signed the blue, shade.

A

Yeah well we'll just call consensus again. Obviously,.

P

Yeah, no, no, no, they just a chorus decided.

N

That you can gas.

K

P

Okay, all right I'll say what is the book about you hey, so they they will do physics.

K

I guess was a dds, a sign optigan. Second, one is probably.

P

Not a tasty tape, hair pills, I need.

N

P

An ash politics in DNS, big or JP eventual did you use to the man. That's chair right! That's! What's.

N

P

Yeah I, don't think it right. That's fine happen. Okay! Sorry! It's right on hey, sir.

A

G

You can ask our.

I

D

What about exercise yeah.

G

As long as little yeah it rasta yeah- oh sorry, so do you want to send email sure.

A

G

Actual parameters and.

K

Signature, oh yeah, so.

D

Do we send an email alert unless you subsequent sort of in there with ya sayin? You know from the consistency.

A

C

You read it yeah.

N

A

Think Jim should be so anti guard because.

O

Otherwise, we will have to the various people. Will the.

E

Visual of it around the.

O

Problem was I needed three more right and they were silent.

A

O