Internet Engineering Task Force RATS Architecture Design Team, 5 May 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: RATS Architecture Design Team, 2020-05-05

Description

RATS Architecture Design Team, 2020-05-05

A

Maybe my participant list: isn't there.

B

Yes, yeah and my cannot to tone phone calls. I says so sorry, I was also on the internet last, so maybe I'm not the best host. Is anyone able to share their WebEx, gets up architecture, side, fuss, I,.

C

Might be able to because.

B

I'm using the stupid tablet due to policy reasons, and if it's me from sharing lately.

A

B

Okay, that's good I, don't know what typically we do here so I might for the time being. I.

D

C

Michael able to here is yet or nothing yet.

B

He doesn't hear us and we don't hear him.

B

That's a new one, so typically it's a.

B

So my question would be: where do we want to start? There was I think a topic we deferred for quite a lot of time since today, and it was theories and I'm not mistaken. He is on the call hi Gary. So we would take that one on because I don't know very good for like five minutes. That.

C

The the question is whether the previous comments have been addressed. Yes, I think, yes, last time, Gary hadn't had a chance to go, make the updates yet, and so we deferred it because of that, and so is it ready or do you want us to defer it again? Well,.

E

The last comment actually got me: you got it got me uh kind of threw me for a loop, but I think it's okay. Can you open it up Hank if you're sharing input the trust this is Dave entering okay, yeah go ahead, go all the way down. We've been kind of fighting over wording and I. Think if we go up just a little bit, I believe Ned sleep, making suggestions, it might be at the bottom. There keep going. You know.

F

We made it just to marry.

E

This one, no, no, he actually bought it. The bulleted item here we go down.

E

You're on the wrong Peter you're on the.

C

Okay, okay, which one should I be looking at implicit trust, yeah that one yeah.

E

C

That makes more sense: okay,.

E

F

There did there.

E

Yeah, so this was rather element and the you know as we go with this, you.

G

E

Tripping on all these things about like what what is it were to trust, have you tired too, and you.

A

Know what is it was.

E

No doesn't really you know: how do you talk to the communication channel? What is the variable? How does the verifier know about this so I think if we're happy with this find a way you know we can find a way to actually put this into the pen. You know the place, the entire PR and put it in just like this, so what I didn't know Dave, because he has the most thoughts on this is. Are you happy with this wording as it stands and I think I think we're kind of done, then.

C

Yeah I mentioned my comment that I think Ned's text here is great and so I would be happy with that.

C

Do you have audio.

C

I, don't know he's not listed, we can't Terrace your feeties can't talk.

B

My connection is also not the best today.

H

Here from there we can you.

C

H

E

I

Microphone was down yeah, so if those words are saying what we people aren't ending, then yeah. Let's do that.

C

So, do you want to take the lead of the PR? Is the question yeah I can.

A

C

You can take your text right here and no Gary. If you want to do you want Ned to create a different PR, you want him to modify this one seems like I think which were just saying a second ago. I thought you said: maybe you want to create a brand new PR and a banner this one or what are you thinking.

E

Whatever, maybe I think we can do it.

F

I'm happy to work on it if you want okay, thanks, Ned, okay, okay, thank you. Thank.

C

You that pretty good, ah let's leave this one open until we get the other one just so it makes it easy to find the text and stuff for people to look back at here, and if you create a new PR, then it sounds like we can close this one without committing it once the other one exists. How does that sound yeah.

E

Not good, it's not good to me. Yeah.

I

What branches is.

C

Implicit trust.

J

C

Use that one you're welcome to use that one or to create another one without was easier. Okay, so.

I

Just making sure it's off the master, yeah.

C

It's awesome, master.

C

You want to create another one trust you or something feel free. If you'd like to work in this branch and just replace the stuff, that's okay, too. Whatever is easier for you.

C

Great great progress on that one thing: I.

B

Can report that there is like a related topic and for sake of Lovins in theory, I would like to give a micro update on this, so we were talking about the UCCS stuff, the Jeremy and John from Google platform or- and yes, we totally agree that there should be some implicit trust section about this in the architecture. So this is full support from GP guys that also on the fire, amongst others. They each draft, for example and I- think that's very, very, very useful good.

B

Having said that, the new poor request, based on my old pull request that Dave found it's in and I think that is something derived from that I. Don't didn't read that in entirely because it was just posted in the night when I was sleeping, I.

C

Was just going to that one like 15 minutes before the call, and so it's chained to a quick pass and it's bunch of grammar and editorial things that I called out in there. Although the main paragraph changes, I thought were a great improvement, but I still found lots of things to come out on so.

B

So net, could you give us a small overview of you and what was the gist you do? You take took from my my PR and incorporate it in this one here piece.

J

B

Issue 86 86.

B

It's your PR about 15 hours,.

I

Yeah I thought I was that you were talking about 55, so I have to remember what I did.

B

The composition, stuff that was awful that took a swing at and then the eighth was we like discussing this honor on a detail layer and then you were created a something was calling I think a revised swing at this based on my stuff.

J

I

Was 55, but maybe this.

C

Is the main replayed to that paragraph so that paragraph with these paragraphs right and so I thought the paragraph changes were quite good. I had some knits in the paragraphs, but I don't know if you want to walk us through this part. Here.

B

J

Is this? Is this.

I

Is out of the the role, composition, master or role composition as the base branches.

C

This is the issue: 55 is the right branch, yeah.

I

Enrolled committing.

C

It into master this is winning it into real composition, you're right.

B

I

So right so basically the just were so so what I? It was starting with Hanks paragraph new paragraph sort of new section six before the previous section six essentially was to was to create the.

I

Context around role, role, hosting roles and role and composing roles did that what I found was we didn't really have a? We didn't have the right turn. We don't. We didn't have the notion of of roles as being identified early on as a thing in the architecture, so going back through it. It's like there's a little bit of reference to roles in the introduction, but it doesn't it didn't really sort of capture this idea that it's a that it's an architectural thing least I didn't think so.

I

So it kind of cleaned that up a little bit and then the terminology will I was a little confused in the terminology because it starts out everything starts out with um you know: a knight Esther isn't an entity which I don't think of it as an entity Eddings as a role of role as being something that's separable from an entity and I know. We've sort of had this conversation a little bit so I kept the entity wording but tried to make it more.

I

More evident that that we're talking about roles here and when we talk about an ax tester and a verifier and a relying party mm-hmm, but wasn't sure people were gonna, have issues with it, but trying to describe it in such a way that it shows the relationships between the the other roles and the messages more. You know work more consistently and then so then, in the context of the paragraph, I try to use Hanks ideas seems like that there were.

I

There were a couple of you, a couple of examples that were being described so I tried to capture those examples in in a few words. Cisely I also took the example that was in the section, describing composite device and moved it down there and and and continued it as a composite device example.

I

There's there are two exam now come as a device example of mixing roles, but it, but that also includes the addition of the verifier and then the other example seem to be one of multiple networks like a a node that is in the confluence of multiple networks. So I kind of rewrote that more concisely.

I

To try to illustrate the notion of multiple roles, so hopefully it's understandable and in the line with what people were making.

B

Using the term role as a tangible term agreed about definitely was to talk about a gesture and define these yes, sir indoors, and such and and then just use these terms and not explicitly highlighting them as rows, which I thought was complicated because at some point we would try to aggregate them, and then we are talking about ephemeral thing. I think this is what countering here right now, because now we have to categorize these terms in through something effectively rolls again and then that you are basically able to collapse them and I.

B

Think probably an entity and yes, I, understand why you don't want to be specific about this I see. I saw Dave, especially Dave's comments about this is probably not a device, hopefully not don't, try to be specific about this. Remove the typically and such and I only said device you to the each focus and and also entity is the thing we can all agree or kinda all agree, the one, but still it's like weird, like having a car architecture without the architecture compose and components being named. It's like very, very hard.

B

So so now we are back at that point again, so we have for my point of view. We have roles and entities, you combine roles of entities, the end, but I think there's still some pushback on using these categories. I think very explicitly to say these. The rolls and these are the entities you use, and now you build your stuff with it. So my question inside this, this monologue is: are we still at that point?

B

Are we still avoiding the terms roles and entities or are we think now that it might be necessary to do to pull them in.

C

Couple points I have no problem with if you're trying to make things we work on, create mentioning a device. I think my comment is in the work. Typically, he said an entity, for example a device or application, or something like that, if you're trying to try it to be real I'm perfectly fine with that. My comment is on the word: typically because in teep and SGX enclaves and things it's not a device, it's an app or a nun clave or a trusted app or something.

C

But it's not a device right may all be typical in different scenario in the different use cases in one use case, that device is typical in a different use case and unclaimed is typical and a different use case and app is typical and so on it. So we could mention multiple. You know, for example, a device or application or unclear wouldn't want to have one or more examples. That's fine!

C

Let's see on your other questions, I'll repeat my opinions. My opinion is that the word entity is perfectly fine when needed in English texts such as in a definition of another term like here, you just trying to use, as you know, something whose evidence, and rather than using something you know an entity just in normal English text. I think it's fine I am against try to define new terms in here, like I would not want to say entity. : I try to come up with the definition of it. I think it's used.

C

Okay in passing, when you don't have a a a term like now something or a thing or whatever an entity is fine. There I think on roles. It's perfectly fun to use that term. I think that was another one of your questions, and then you asked a another one which I didn't like and I forget whatever term. What did you use? Oh no I've coming down empty role,.

B

C

B

I said the catch-all term: these are all architectural which we don't have to define this attraction, but this bucket term was circling back to this yeah. We had actors, ones, I, don't like actors. So the s GX is an actor in artistic I.

C

Think roles is a fine term to use like here. You know: I saw like Ned's title here: I think is fine. I hear he capitalized roles, I think my preference would be for lowercase roles there so that you don't have to define a term in here, meaning I. Think if it's used in passing a couple times I, don't think it has to be for more value.

C

I I think your point is: can we call it a test or a verifier and or sure can they call those roles and I'm fine with that I've never had a problem with that. So.

B

Say that roads are there, there are things and we host them on.

B

C

Hosting or you compose them I think was Ned's other term. Here, I thought you had both.

I

In the title, because I wasn't sure yeah.

C

I don't know if host is the right term. Maybe it's just you know: composing roles or role compositions or something cuz, I. Think composition is a good word and I thought that you used composition somewhere in your text here, Ned.

C

Combine accommodations is good, I thought you smoked, Levin did ya.

C

Hear you used hosted.

C

But you know combined in that would be in. Is it perhaps down.

I

I use performed yeah.

B

Roles reside on an actor, so that is, that is very common terminology. You have an actor, that's the thing: the sjx s essentially removed actor because again duty highlighted, we don't really want to reduce neutrons, we can use an embossing and then you use establish terms seems.

J

Like Dave's use of antique.

I

Includes things like you know, in an ontology your role as a thing right, everything's, a thing in an ontology.

B

By diffusion, everything almost.

J

Seems like it's, the entity is the equivalent of thing.

C

Right now, I really like I'll, be detected. You have here Ned, you can see. Usually I have zero comments on it, which means I, really like it that you'd be applied if I get to the end and I hated comments after that and I skip a large section new text. That means this is good text, so good job.

J

The question of do we do what's our position on uppercase role,.

C

We use uppercase for things that appear in the terminology section, so.

B

My take this was uppercase role is not not to Dave, so we could introduce a small sentence. That says there are Reds roads or there are roles here. We don't have to call the Reds rows, because I think it's dumb I'll.

C

Call the station rules you could but I, don't think it's necessary. I. Think lowercase roles is perfectly readable to the normal reader, without definition in all the context that I looked at I think it's unambiguous and perfectly understandable. Okay,.

K

C

Because right here you define it in context right right context, and so you don't need it to be capitalized. It's aren't here.

J

Because with role messages for casserole messages, I hate.

B

C

Master, we already have conceptual messages as the section heading, and so that's why I said I change this back to match. What's in master right now, because this PR I don't know if the I hadn't noticed what I made these comments- and this is off of issue 55, not off master, and so I was comparing it to master and in master it's always called conceptual messages.

C

J

I

C

Opinion is readable, but that's me.

B

Let us call it conceptual messages. We literally have a section about this, so I think it's fine to use conception, but it's not it's not rolling off the tongue, but neither does raw messages so I'm.

J

Fine with that.

B

Okay, so rolls are more prominent buts. The lowercase entities, as the used example, should not be to occur. Yes,.

C

You can see here in I, guess: issue 35, that's your branch right now everything! That's my branch, okay yeah! So you can see here rolls used to be location, I liked it lower case. Your wording changes fine, but I would keep it lower case. The.

B

Red one is from mine, the green one is from yeah yeah.

C

Sorry, you're all composition that had roses lower case here. I'm just reading keep the lower case, as it was in real composition, conceptual messages, otherwise the wording changes accurate. That's my theme and.

I

Then the the the only typically that you want to change is the one related to the tester.

C

Typically, it's like I didn't come in I. Think that's fine, I think you know the here's are typically that one's good, there's typically I, think that's good, because this typically is they're all about the endorser role right saying an endorser is typically a manufacturer. I think that's fine and a relying party, you said, is typically an application unless somebody else speaks up. I think that's true in the various use cases that we have in the use cases list this.

C

One I hadn't worked a problem with, though introducing a term service provider, which means different things to different people and I would prefer to avoid that.

J

Any suggestions.

C

In the original text was a more ambiguous, isn't it some people, the service provider, is an ISP, some people, it's a description service. You subscribe to that's completely separate from your network service and be various other things out of context. So.

I

Right so I was trying to find I was trying to be consistent in providing intuition and I agree that find anything better than.

C

I

C

Intel arm and Microsoft are all things that have verifiers today and I'm sure there's a much longer list as we look across data use cases.

J

So so just know those are consensus to just not have a into intuition just.

L

C

Provider, typically, a service yeah.

L

Typically, a service here.

B

But yeah, okay, but the service is fine. I think it is such a soft term that you can consider this.

C

There's nothing typically and that already had a such as so you say you upgraded from an example, a typical I'm, fine with that and I think. That's all the typical ease in here yeah.

C

Okay, is there anything else, a highlighter you wanted to talk about I. Think we've now covered pretty much all of my sentences that aren't simple editorial ones. We probably don't have to discuss.

B

This we merge this I haven't.

L

I have a bunch of comments. I'm I'm, making.

B

L

Was just gonna I'm just writing them here, but one of them is avoid the term trustworthy, because I don't just think that that's just fraud.

B

So the child uses the word, trustworthy, I, think 12 times or something it's literally. The core of the work doesn't.

L

Mean we have to use it.

C

What's the problem of the word trustworthy trustworthy with respect to a particular point of view, which is in this case it would be the the verifiers point of view, but what's the problem of the term to.

L

Me, the relying party is going to make a decision to provide service or not provide service or provide some level of service, or not quite some level of service, so I would always just put it. In those terms, the relying party decides whether it's going to provide the service or not, based on the the attestation result. The.

C

L

Know what is relying.

C

Party, it's about the verifier, the verifier does something: can you use trustworthy or a flower to something that the verifier does or what's the problem with the end of air fire context.

C

Buying party not disagreeing with that, but just the context here is in the context of the verifier, the verifier.

L

Uses evidence and endorsements to produce the attestation result. It doesn't make any decision about whether it's trustworthy or not. It just produces the attestation result. I.

C

Don't think that's true for a verifier, it's at least for some purifiers. Some of them will produce edict and explicitly negative attestation result to say this thing is not trustworthy because it contains the following. You know because it contains vulnerabilities or because it's has a time stamp, that's too old or whatever, and so they may say this should never be trusted. As you know, it's a signed statement from the verifier. There are verifiers that work that way.

K

That trust where the interface is bad right, because it really isn't about trust. It's about whether or not the claims the evidence support are sufficiently substantiative for the use.

B

Trust position is a prove effect by this. What you just said, so trust worthiness is not trust. It's different things.

I

The claims are intended to be claims about trustworthiness and therefore I.

L

Don't think so the claims are not about trustworthiness. The GPS coordinate is is not about trustworthiness. It's oh, it's a GPS coordinate. Some would argue that. That's not a great example.

H

C

There are cases where GPS is used for trustworthiness, for example, if the GPS says the FET machine is no longer in my data center, I don't consider it trustworthy because it's been stolen or is no longer in a secure location and GPS can absolutely be about trustworthiness. I can be used in trustworthiness, decisions right, a secure location versus not in a secure location, anymore.

B

I

Don't think that we're trying to design a telemetry system is the angry I mean.

C

I personally, do not have any problem with using that weren't trustworthy in the verifier context. I agree that it's not appropriate. That is it's it's better ways to phrase stuff in a relying party context, but in a verifier context, I think it's fine! That's why I'm trying to figure out if somebody can convince me why it's bad there, because I'm open to that self. We have a better suggestion, but it doesn't seem bad to as long as it's not just trustworthy.

C

What do you mean so if.

L

You say if it is trustworthy, it's a binary decision. Verifier doesn't produce a binary result. A verifier produces a complicated adaptation result that needs further evaluation by the relying party in general. That's true, yeah I agree. You've got a I mean to me. This whole thing, the only the only person that the only thing that actually gets to make any kind of a binary decision, and yes or no in this, the whole architecture is the wrong party, and even there it's not a binary decision. Often it'll be like okay yeah. You can have.

L

Actually you can have this amount of this size transaction or that size transaction or something we.

I

Do eat a trustworthy to be binary I, don't.

C

But I am noticing I think I'm, getting somebody else's point now the phrase that I have highlighted right now and move it up a bit here. The phrase that I have highlighted right now I do realize is problematic in one sense, and maybe if we fix that sense, it'll fix people's problems with trust. Already here, that's not mine, so this is talking about. The evidence has to be appraised such as when deciding whether the ax tester is authorized to perform some operation right.

C

This decision here is done by a relying party yeah, the determining whether the evidence is considered trustworthy is something that's done by a verifier okay. Now, of course, this is just a such a such as wind and verify when it remind party needs to decide stuff. This phrase is perhaps problematic in terms of an ax tester as phrased and I.

C

Don't know what that better way to fix it is it's not completely broken in my view, but it's probably maybe there's a better suggestion as to how to word this, and maybe their suggestion would just happen to avoid this issue. I, don't know so have to think about that I think.

L

She should do is just the Vera doose's.

I

Describing the ax tester.

A

Okay, I was miss reading that sentence. Well, so David what you just described. You said that you know it is considered trustworthy. You used it referring to the evidence. I was originally reading it as it referring to the attest. Are they.

C

Written you look at the read whether the entity is considered trustworthy, so this is about whether the attest or is considered trustworthy and whether the ax tester is authorized to perform some operations. That's why, when I said this, it right here I expanded it to match what the original text that you see. That's my point down here, but.

L

The relying party may vary its transaction size based on some characteristic of the stir, the attesa. Maybe it's certified at a certain level, and it gives us this. So it's not a binary decision about the adjuster either.

B

If you have to highlight here that the the you all trust roads, yeah verifier, to be consumer of its evidence, relying party trust a producer of the iteration result, you have endorsements here, you trust signatures. That's all trust.

B

Signatures are not about trust. You literally trying to tell me that signatures are not a trust relationship.

L

Signatures tell you that the thing you know has integrity to the to some to some degree based on what you know about it, but the problem with the term Trust is that entrust and trustworthy is it's always just by the nature of English language. A binary thing.

B

I

Think that's true.

B

Again, a phrase.

C

That I thought up here that somewhere else, I I could be misremembering, but I thought it was in at least one poor request. I don't know semester or not, and the extent to which is trustworthy right in which is the phrase that says it's a spectrum. It's not binary. I was.

B

Only talking about trust, so you decide, you want to believe things that are signed, but the signer by the signer. Then you put trust into the signer and Trust into the signature. That's a decision, then you have content of the thing that assigned that is not binary most certainly and then content of the thing as assigned is about trustworthiness and then the verifier most certainly makes an assessment of the trustworthiness and the relying party still cannot take that into account or can take that into account either the trust severe fire.

B

It can still make its own decision about how to processing these attach station results, but.

L

I agree that there is the result of a signature. Verification operation is, is a binary, is binary, there's no degree in the app so but sewing I. Think.

I

The so I generally dislike this definition of a tester, because it's trying to define a tester from the perspective of both the verifier and the relying party, and it ought to describe the a tester from the perspective of the ax tester. Yes,.

L

I

I think it's orthogonal to this pull request. What what are what the definition of you know the wording of the definition of role? Maybe we can we've been open, a separate issue that to revisit our definition of a tester, maybe the other roles.

B

Exchange the it's in this line, 2 1 to 6 year with the tester yeah.

C

That's what I said my comment down here is the the Green is not correct, because Green changed it to refer to evidence right being the the last noun or the the reference and reverting it would go back to the entity or the ax tester to be even work there yeah. But that's what I comment it down here right.

I

So I would I would advocate for but just revisiting the definition to try and not do any.

C

Different this one defining.

I

We're trying to define both the verifier in their line party in the context of defining the tester which yeah maybe there's no other way to do it. I don't know, but it seems odd to me so.

K

One other common here is that the the word determinant is too definitive here for trustworthy. It's it's really.

E

The evidence is only.

K

Supporting a determination of it right, you can't ever determine trust for this. All you do is push trust around. That's.

I

Why we use appraise and appraisal policy I, don't know why we don't use that word in here as well.

C

Because this is trying to say that there's a process here without saying how it's done think of phrasal policies in a different by definition, any product on the page, I know.

I

But at some point the all the definitions become intertwined, but if you choose different words, then people have issues with the other words. So if we've found words that we agree on, we should use those words and not use other words. I think.

L

It's quite fine for the definitions to refer to each other, I mean they really do have deep relationships.

C

Right now, my preference for this PR as opposed to a future one that Ned was suggesting, is that the red text than fixing this, in other words the the whether trustworthy, sounds binary and to determine the extent to which is trustworthy, I, would go ahead and make that change here. So I think that isn't, if route went to say it's not binary or it is in general, not binary, but other than that I can live with the rest of the red text. Until the IPR comes comment,.

B

About to determine and to substitute determination, I think it's up to, but it would address also another concern. So maybe we can also do that here.

K

Would that be okay, I'm.

C

Not sure whatever I understand was.

B

The previous comment on determine to determine, if that comment repeated, then maybe we can incorporate that here. That was that.

K

Was my comment I was playing with determine was to definitively the whole purpose of the attestation is to support a determination. The.

I

Determination, your use of the word determine our determination is in the context of the relying party making the determination.

I

This line is trying to apply that determined in the context of the verifier. We said we would use appraised evidence right.

K

A single claim, or is it a body of claims, that's.

I

A body claims yeah.

K

So if it is the body of claims, then your that determination is inference based on what those claims are yeah. So what is in furball will depend on the set of claims that are there.

I

Is that a better word to infer whether the entity is considered trustworthy here.

C

How do you like the green text on the screen right now.

C

If everybody like just I'm, going to go ahead and commit this into Ned's branch, and just we can make some progress here and then Ned can go through the other minor editorial stuff, then we would Merck I'm gonna go ahead and convert this right now, a set.

L

Of review comments I'm submitting I'd like it, you know pull requests needed like a few days, at least before you merge them. So if we can actually process, can he.

C

Keep in mind that this right now is not emergent to master. This is emergent composition, so we.

I

Still, we still have role composition that you can comment on trying.

L

To be productive,.

C

Is there a place that you have to look at right now that you submit your comments? I'm.

L

Trying to do a review here, but everything is moving so fast I mean I.

G

L

G

The flies so I.

L

Haven't had like an hour since I woke up to look at it, so you mean you do a pull request. It's you typically I mean my pull requests it out there for months with people, people looking at them, so I'm.

C

Really on the ball here.

B

This is my pull request and I. Do not mind this being put into my appalling face this citizens, just okay,.

C

So we have about nine minutes left. My question is two more. Would you prefer us to discuss any of your comments Laura and see? Would you prefer us to go into a different issue and come back to this next week.

L

I'm happy, let me see if I got a check here, see if I got.

M

What was the resolution around line? 624 both rows next.

C

Week, I'm just gonna try to answer what was it. 624 was a question for.

M

Ya there, what.

C

Was the conclusion conceptual.

M

Messy you're gonna we're gonna we're gonna. We're gonna go with your suggestion to conceptual messages, because.

C

That's what's already in master right. Roll messages was some thing that was only in rural composition. It's not master, and people were agreeing to. The conceptual messages was at least as good I thought. There was a consensus on that. So we're on the call, probably when, when Michael you didn't have any ability Harris during that yeah I had.

M

Yeah I wanted with audio but not not green, for a while.

B

Messages to conceptual messages, as they are in our composition, Michael.

C

Do you have screen now?

C

Can you see.

M

Okay, yeah anyway, I had opened another another tab. So I think that you had a bunch of editorials which I put into the batch, which we should commit and but that for which one for the.

C

486, okay, yeah Nick, yep, yeah, I'm gonna. Let Ned do that because I didn't touch every line so, for example, the roll messages and conceptual messages are in there a bunch of times that per request, and you need to be a bulk searching a place, though so I wasn't going to merge it right now. It's gonna, let Ned barge it into into the Royal Composition branch: okay, with people, okay,.

B

C

So I just being explicit Ned in the issue, 55 range there's a bunch of uses of roll messages. I only commented on the first line and it appears like six times or something like that, and so you just do a global search and replace and then go ahead and merge once you've aggressive editor. All things.

C

Merge into royal compositions, which is still in the master right, we'll review it on their quote: roll compositions branch into master per request. So we can review that part next week. Does that sound okay issue, fifty-five merged into brawl compositions? And then next week we can review the discs between the enroll composition, semester.

J

L

We're doing a lot of work on defining terms here in this poll request so which is not really roll roll composition or I guess it is I would never mind. Okay, yeah.

C

We had six minutes left. What do we want to do in the last six minutes? I would love to get people's opinion on the term. The Strummer interim that I threw into 85, which we haven't had discussion about. Yet I would love to take five minutes to just get people's opinions on there to see if they like it or they hate it or they have a better suggestion, a different suggestion, because that would be my own preference, but could.

B

You open it up so that this day.

C

Know the I'm gonna go to the exchange. You can see Hank, we are looking for a better term than you fee by us. Rumors, there's a bunch of places that said you fee, BIOS firmware and so the very first a testing environment test to ensure the integrity of the you fee, BIOS firmware whatever that initially boots up a composite device. Okay and so I tried to come up with a term for that first thing: that's above the hardware. Okay I mean not. You have Robert, okay and so I tried.

C

My suggestion was the first mutable environment example BIOS our firmware. It's.

B

C

D

C

E

E

This is get real, quick, sorry, I mean that's a breaking, not I did look through that I, like the text. I know what you have what you have intended. We just you know in the IOT space, so you know with heavily allowable code. We as the attend. The rue des trust may only be attesting something that's essentially our what we would consider Hardware function so like, for instance, the you know, there's one-time programmable, there's one-time programmable memory.

E

It may just be reporting on that, so it wouldn't be mutable from that perspective that a programmable part of a important you know a programmable part of the environment. The target environment would be so if you want to cover that time, if you don't even you think it's just going to complicate things, we can leave that off but decided to you so start a question.

C

For your Gary, this isn't a section about layer. Data station is the case that you're talking about a case uses layer data station, or is it only a one, a testing one target environment case.

E

It can be layer depending on it can be layered if the root of trust, all it does. Is it just uh it reports as reports a hardware state, and then it just triggers the it triggers a power on above a subsystem, a target environment, but that target environment you know is basically all long code. It's not really it. It goes ahead and if it needs, if it's in its verified its own patch, therefore it's just triggered a power on of the of the target environment.

E

The target environment itself, though, is the one that actually verifies it if it verifies whatever it is the mutable aspect of its code, that's running so I can put this in a comment on there, but I mean it's a you know it's actually when I'm describing is actually sometimes you see it a lot in the IOT space. You don't see it, you don't see it really in the PC space. So.

I

But this the section is in the context of providing intuition in the form of an example and the example is a PC BIOS example.

E

That's fine! That's why I keep it. If we want to say this is clearly for PC BIOS and that's fine as it is made it, then that could point the intent that we can do it. We can say the other. The other point in the other corner cases that I'm thinking about you know, maybe maybe you don't need a lead and if.

B

You could highlight a prominent French case you, it's not French legal of course, so a prominent use case for it from your side. I think that's helpful in an issue tracker, so they lose this and can make a deliberate decision on ok. This is you don't have to elaborate on this. It's implicit enough I think it's very important to have this. The small things are less mutable than you think they are here.

C

You would be ok if we choose to not cover it, but I think that the case you're talking about us were the first two layers are both immutable, as opposed to the first layer being immutable in the second layer being mutable. Alright, if I understand your point, Sam yeah, yeah.

E

C

As long as it's just an example than it is fine, because the whole point is used to describe how layered works and how it actually Maps, it's just an example. If it's.

L

An examples stick with you feed BIOS. Then you know it's really an example. They're not trying for something else.

C

Probably good firmware can beat mute. You can flash firmware on stuff, so it's all firmware.

L

C

Gonna use this term as long as it's kept clear that this is just an example and there other cases you could have you could have layered attestation with like five layers of all immutable stuff. I. Don't know why you do that, but you could yeah.

L

Because you can beautifully, isn't it isn't critical, hear.

I

The term mutable gets us into semantic issues. We cut yeah I, don't like it. What at the end of the day, mutability is controllable and, depending on your perspective and where you are in the supply chain, either is or isn't mutable and it was.

C

Mutable ever by anybody right, that was the intent here. It's.

I

Never the case yeah, that's almost no really case. What's.

C

Never the case what I meant is, if anybody can know.

C

J

Word appointing is the philosophical thing that doesn't exist, a.

C

Better suggestion I heard several people that say that they like this. What do.

L

You feel by us I think, that's better.

C

The term was UV / BIOS, flash firmware yeah.

I

Understood: okay, firmware right just as long as this! In the context of.

C

Words we're trying to get.

I

C

Of this we're trying to get rid of this to do we're looking for a better term, then, and so we're trying to get rid of this text, we're trying to delete those three lines. What do we do? Do this same.

L

I'm saying that I think that is a determinate you're using is good, because then it is an example. Don't get rid of it.

I

You can, which is like the you know, PC firmware, for example, UEFI BIOS, blah blah blah. You know, I mean it's like yeah. It's not that hard of a problem in the context of an example of a piece scenario.

C

So, who wasn't the spoke correctly for net? Was that Lawrence I'm, not yeah, I, think.

L

C

Lawrence, can you just explain what you think the text should say in like in these few lines, which is that you like red or green or something else well.

L

I just haven't I haven't read the text in detail, so I'm just I'm, reacting and I'm. Let me say: first minimal environment you're trying to go abstract, you're you're. Then you start getting it, calling it to question. What's mutable, what's not mutable and all kinds of it looks like it's starting to look like a definition. Well, if you say yuffie slashed by a slash framework, it's clearly an example and just an example. It's very specific and and clearly the world it doesn't all use it use you use UV and from work clearly other.

L

There are other things out there. So you're just trying to be an example state as a state as an example, all.

C

Right so we're over time, but I understand you saying you like the red text, better, correct, yeah, generally, yes, okay, I, think we're out of times, but- and so I just wanted to highlight this. To get your first impression, it's feel free to go through and review this offline I just want to highlight it and see whether it was general likes your general dislikes on this and so I. Don't know that we came to a conclusion and that's fine.

C

We can come back to this next week and so I just want to introduce it, and let people hear what the people were. Others were thinking. Oh, should we pause here and resume in a week.

F

C

Other people have meetings to get out too. So let's go ahead and stop sharing here, I, don't own the call, but I was sharing so okay.

C

Thing thanks Paul for house.