Internet Engineering Task Force 97, 15 Nov 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF97-TRANS-20161115-0930

Description

TRANS meeting session at IETF97
2016/11/15 0930

A

Then that gives the log operators the ability to make that policy decision. But in the absence that that would be the only blocker for the best document. And then I think peter has some stuff to say.

B

Okay, let me try and hit the the new button so Peter you can now speak I! Think! Yes,.

C

Can you hear me we do know ah so I think it. You know, there's the: how do we technically do this and I think the other half is understand you know, should we do this? I mean I, know, there's been a lot of question about what part should be redacted, and you know because this is also the previous slide suggests. This is just looking at subject alternative names and be NSA's, but it's also been brought up. You know that subject names frequently end up with your personal information in them and I.

C

Think that that saying that needs direct addressing on this, because you know yes, logs might now be able to reject sorry might be able to now reject serfs moving to should, but not good that they should reject every sir just because it has what appears to be a personal name in this earth that doesn't actually solve the kind of core problem of how do we get to a you know, world of sikhi everywhere.

D

So, with the periodicity correctly you're saying that the like policy change from a master shoot by itself will not be enough to address the issue of of personal certificates.

C

Well, I'm saying it: it half addresses it because it allows the blog to reject the request to log a personal cert, but it doesn't I mean if the. If the objective is to log all search that are reasonably usable on websites, I mean what do we do when when there is a website that has say a first name and last name in a certificate.

A

Peter, would it be fair to say that you believe that the must to should change? Is the right change to be made for the BIST document and that the scenarios relative to personal, serfs and redaction are something that should be covered in the redaction specification independently.

C

Yes, I think that's a way that the muster should a lot of 1660.

E

C

This to give the opportunity out, and then we still I think, is a working groups off to address the other half of the problem.

A

Do you have other scenarios that you can think of as examples that you think should be covered in that this document that we should consider as scenarios well.

C

I think the other one that has there's been suggestions are changing. The best documents to a movie, techno constrained, option arm or my beloved section or I think the concern there is there is also even at the base domain. Their third question is: do we handle you know the buying the binding of just knowing that Ed short of a cert is to a different domain can be automatic, especially in the kind of enterprise CA scenario or, however, you want to put it because simply knowing that you know, example, corpse CA is issued for new thing.

C

Calm can actually, you know, get privacy if you origins a privacy to incorporate secrecy issue, it sort of almost conclude chairs, but it.

A

Thank you that that's a great example of a scenario that that really should be considered in the redaction work.

B

Okay, so so then I think the last issue we have in relation to the bits document and pushing it into ITF last call would be the historic fetching of sths, and so is there. Anyone who thinks that the addition of this API would be a problem, and they said no one who really wants it. So.

D

I, as I mentioned, I think it would be like a nice optional API, but you would be okay, but I like, if there was that come with monitoring api, a document, then it would have liked that document will do in a much better place for its api. So.

B

So you don't think it needs a change in the biz document. Hey squirt.

A

I've, it looks like Rob straddling isn't in the media meeting here, but I've spoken with him and I know that he forcier TSH. It would like to see a REST API that would be standardized for services like that, and my personal opinion again. This is Ryan. Hurst is that that is the this particular API is an example of something that belongs inside of a broader query: API. Instead of just a bolt on to the log infrastructure.

B

Okay, so it seems most people want, is not to be in a biz document itself and in the the new log monitoring API. That was.

A

B

I see lots of nothing and nobody rushing to the mic. Okay, okay, so that would mean I think that once you update the section 51, we can put the document into ITF last call.

F

And in the meantime, if somebody wants to draft at the death, the API, okay.

B

um So then we get to do the two new items, new work that were on the agenda. One is the log monitoring, API I'm, not sure, if just anyone here that wants to talk about that, but otherwise we can just keep that discussion on them on the list for now, because there definitely seems to be an interest, and the other item is the expect CT TLS header, which will be discussed later in the HTTP piece working group on thursday.

B

So that's not something we need to discuss here. I think.

B

Go ahead, yeah.

D

Around me, sir, so just one more thing for this I've started the reference implementation, which is currently at my personal, like github repo, and there is a link I think inside 11 of the presentation it.

D

So it only does like two things right now, which is like adding a chain and then providing an ST age, but it all does it always like with 66 to these data structures, so I I would appreciate if people had a look, I'm particularly looking for fightin reviewers for the code, so I can merge it upstream back to the back today, like main open source, CT, repo.

B

Okay, so so, jumping back to to the other documentary self open, just a threat analysis document, which we have two people that are sort of conflicting their opinions and the document has been stuck for. You know half a year now. So what we're really looking at this? Maybe someone who's willing to become an editor on the document so that we can just fix the last issues about the duel CA attack. I move this document into working group last call. So is there anyone who would like to volunteer to do that task?

B

That's what I was afraid of yeah.

F

I mean this is pretty straightforward. It's just basically one paragraph. It should be pretty quick. We will clear the way for you, in terms of you know, process in dealing with people and so on. We just need somebody to UM to bang this out. I.

D

Run again so yes, I'll, be I'll, be happy to take on the changes necessary and while I'm doing that, I'll make sure it reflects. What beast currently has.

F

Okay great, thank you very much that.

B

It would be really good to get this document out, because it's basically almost ready right.

F

I mean one of the problems we've been. Having is that we've been getting absolutely no feedback from the mailing list when we raise this question and it's made it impossible to to move on. um So if anybody does have any opinions idea that you know time to raise them here, if you like, and if not we'll just you know, appreciate Aaron having volunteered to to Daniel's out and and then I'm really shopping for that one. So as soon as it's done, I'll get it a write-up Dunham through the process.

G

This is general Khan Gilmore, so I feel a little bit responsible for this impasse and erase.

C

Me initiate attack.

G

um I tried to do a review of the discussion in the back and forth and there's just a ton of texts. Yeah.

F

G

There's walls of text in multiple different directions and there's complaints in different directions and I found myself stymied by you know it wasn't even like everyone was agreeing on what they were trying to do in the discussion, so I kind of threw out my hands and I was like I like. I can't I don't know how to unpick all of that um in terms of specifically describing the duel CA attack I felt like the text that was there that was describing the attack was correct.

B

So yeah, so you mean it to fit text from David that isn't in the document or the checks.

F

It isn't enough I.

G

Believe that the Texas in the latest version of the document it that, but this is what I mean the discussion around- that text was like well there's this other text, thats related to the duel, CA attack that isn't the fault of CT. That shouldn't be present there and I write is this. Is this? Is my memory of it?

G

I haven't looked at it recently because I don't want to reach Rama ties myself, but but when I, when I believe it's the most recent version, I believe that the Texas in there does actually describe the duel, CA attack, that least stuff. That is talking about the duel CA attack. Now, whether that's the fault of CT or not, or whether that's an acct can fix or not. I don't know, I don't believe that CT actually resolves that issue.

G

It's arguably it's an issue with the with the our inability to effectively revoke but there's just like so many different moving parts that it makes it very difficult to come in and say yes, this is good because the arguments are like well. This there's other pieces that are associated with it. That aren't actually, maybe necessarily appropriate and I. Don't know. I didn't know how to resolve any of yeah.

F

I mean basically, I think the right way to break this down is to delete all previous state completely. I take a look at the current document. Take a look at David's response to the current document and don't look at anything prior to that that make sense I.

G

Think that makes sense, although there may be multiple replies, I think there's a back and forth set of replies. David's.

F

Post tend to be pretty self-contained yeah.

B

Yeah, you can only read the David's latest reply and the art the draft text. Those two bits of text should be the only thing you need. Okay,.

G

But there's still the I mean in David's reply, I believe there is also still a set of questions that aren't about whether the duel CA attack is sufficiently described. It's also about whether there's additional text that isn't about me.

B

Ray so feel free to ignore that part for your review of the dog. The text of the degrassi.

H

B

H

It's all talk, my um as those. If you look at recent chromium and mozilla mailing security mailing lists, you know what is two different cas is pretty much like. What's an np-hard kind of problem to solve these days, so I think describing the attack, we can't do it if only because we can't determine if this CA is actually different under different administrative control than the other CA and the MOOC last call.

G

This is dkg I'm, not actually sure that administered control is relevant to the doula CA attack right I mean I. The dual see attack is that there's two different particular public keys that are associated with two different cas, whether they're, administratively different or not, is irrelevant to the particular attack.

I

G

Right for that so you're saying that the attack is actually much more relevant today, simply because there are multiple things that are under control of one party yeah.

B

Okay, so does anybody have any other business.

E

Which I live all yet then you'll you'll get emails. That is what you get sooo much 48, where you ever do into the last station in the ER and then others boobs on ya, ready just a techno pop okay, that's triggered.

I

Hundreds of you already did because you're.

E

Ready to divide within you need differences between SI pensa q to end this, and I will call that it plus the latest game to catch any. We found a man / system, so thank you to do my job.

E

J

Thanks to cover for other groups,.