Internet Engineering Task Force Interim Meetings, 14 Jul 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF-CBOR-20210714-1400

Description

CBOR interim meeting2021/07/14 1400

A

Yeah, that's probably a good.

A

B

There are a few people over in the in the webex for who might wait to migrate over yeah yeah other than that. It looks like a very small group for today.

A

It looks like the connection breaks down and each time the connection breaks down. Of course, the screen sharing breaks down so.

B

A

So, let's see how long that, actually.

A

B

And just out of curiosity did any of you um put the meat echo captures attendance note into the into the minute, because I don't see, who put that there might might have been, might have been barry, but he can't be here today. So I can't ask him right away.

B

I asked around at the at the chairs mailing list but didn't get a feedback there yet, but if anyone can confirm that meet echo does capture the tenants, it's fine with me and we don't have to do the blue sheets in the code. Dmd.

B

B

B

Okay, I'm being a few minutes in and not having seen any response from scott who might still be in meet in webex, but I can't really reach him there.

B

I suggest we get started so welcome to this last seaboard interim before the next itf. We have a very short agenda for today. That mainly consists of document status and the topic of basically the topic of.

B

How do we ensure that the list of tags um stays usable which comes out from the earliest from the from the other documents?

B

um Is there anything anyone wants to announce to bring up now for other business or for other points in the agenda.

B

I'm hearing none I'd like to give a brief update on where the documents that are currently in the working group are. If they have changed recently, one is cddl control which wasn't an extended working group last call for lack of for lack of document reviews hank just provided one today, so, which is largely positive with just a few knits and and texture remarks that I'm positive can be worked in easily, um with which I'd conclude I'll conclude. The working group last call positively so carson I'd.

B

Ask you to work in anything that you would like to pick out of that um out of that document. Out of that review into the document, and um when that is done and uploaded we can um we can. We can forward this out of the working group.

A

So, do you want to to see the changes in your shepard review, or do you want to do a shepherd review now.

B

um I'll start I'll start work I'll start doing the shepherd review I'll start on the shepherd review I'll work them in as they come.

B

If you, um if, if that happens before um during the during the freeze, I can even look at the document as it is being in the editor copy and push the button on the shepard review on being done. When the documented shepherd review reflects.

A

Good, my plan is to to put in some changes for the comments we have by the end of this week, but I can only submit it a week later.

A

So maybe we can collect some comments on on the changes that we will have by the end of the week and make sure that the version we submit on on the 19th or so is uh ready to go actually not 1926..

A

It's ready to go the window.

B

Goes open again.

A

B

Yep: okay: let's do that.

B

um The two documents network, addresses and file magic are both in a very similar shape. Michael thanks for uploading. The latest versions before the cutoff is: are they still from your point of view and a change where you don't see anything added to that that that would need to be done further.

C

I think I'm at that point where I don't see anything else that we can do to them. That would make them better or.

B

Worse um well then, I uh I started working I'll start. A working group last call on them um for for the duration of the of for the duration until the itf, um unless any comments come up here now on the document, but I don't expect any, but that sounds.

C

A

Okay, so gustan seems to have some some audio problems, um uh so the the reason I'm uh showing my slides uh the whole time here is that for some reason my computer takes a long time to actually start screen sharing. um So a discussion that came up was in the context of the network. Addresses tags was whether we would say anything about the existing uh tags that do something with network addresses and um that kind of brought up the more general issue.

A

uh What do we actually have to do in the long term to keep the tag registration mechanism?

A

A viable part of of the sibo ecosystem and tags are pretty important because they are the main extension point. um We also have the simple values, but those have only very specific usages and actually so far uh we haven't allocated anything uh there yet which might change with uh seabor pact. But tags really are used a lot and they are a part of the batteries included approach of uh cbo.

A

So you are likely to already find a tag for what you're trying to do at least that's what most people uh find and if they don't find it well, then they can go ahead and register one and it's a pretty easy uh process. So actually um some of the ranges are under. First come first served and it's a vast space, so we are never going to run out of the entire space. But of course, short tanks are better than long takes.

A

So uh we have to be a little bit careful about the the space with the short tanks.

A

So the designated expert only sees the the first 32k, um the the fcfs space actually began at 256 with 70 49 and with eight nine four nine. We decided to expand this a little bit, uh so we have a little bit more control over the oneplus 2 uh byte tag space as well, so uh just to to have a little bit of a numeric foundation here. um The the scars part here is the oneplus 0 part where 7049 allocated nine tags, since uh three more have been allocated and there are 12 uh free.

A

So we we have gone through half the space here uh in one plus one uh slightly less less efficient, seven tags were allocated with 70 49 57 tags have been registered since and about 170 are free in the oneplus 2 space 43 have been registered since 1749, and we have a lot to free and the oneplus 4 and oneplus 8 space are essentially abundant, so we don't really have to care about that.

A

um But that's, of course not the only uh problem we not only have to avoid running out. We also have to make sure that the tank space is accessible and isn't such a disaster as the port number space, where people typically don't find uh ports, they they might want to use. That's really a rare accident if that port isn't already uh widely used like port 80 or 25.

A

Nobody looks into that registry, so we have to think about evolving that space as something that that is accessible to developers, and so what happened in the first eight years uh was that um we had this initial land rush when when 749 came out, so we had tax, for instance, for the pearl community.

A

Later the hesker community came in and so on, and then, after this initial land rush, we had some some pretty steady growth, so people registered something but also standards registered, something like cozy, cut and uh other internet drafts, which are not yet standards like the time tag which is being used in the haskell, and there also have been a few efforts uh actually generating some structure there.

A

So 8746 is for array, shaped um structures, and that is a nicely laid out, a set of 24 tanks that can be used for this, and we are trying to do something similar with the map like uh tags. So this is where we are, and I don't think that there's anything that's completely going wrong here. Of course, not not all tanks that have been allocated in particular in fcfs are examples of wonderful, great design, but that's not necessary.

A

If somebody has a specific need, they can just come in register their tag and and get their work done. We also have had very little problems with squatting so far, uh so take 42 was the the only tag where we had uh squatting that we didn't notice when we allocated take42 to yang zebo, and so we had to change the yang siwa to avoid the conflict between gangster and published open source software.

A

Okay, so uh do we need to do any clean up so that there are tags that replace older ones, and I think the network address case is the first real case that that is getting close to working. Our last call here, uh some tags complement others. So when we defined tech 37 in 2013, we really just didn't think about the fact that mime is mostly binary, so somebody went ahead and and uh actually registered 257 in the the uh first come first served space at the time.

A

It was first come faster, so we now have a different tag for binary in mind, which really doesn't make a lot of sense.

A

uh It would have been much more reasonable to have a single tag for for both, uh so that that also opens the question whether we could have fixed the tag 37, but that's of course uh difficult in a registry situation, and then we have other tags that simply extends previous one, so take two to 68 and just to 70 are slightly more flexible numbers than the one that have take four four five and something else I forget um so that that's again a question could we have extended the the already registered one, but we run into uh a problem uh here.

A

So if somebody comes in and says, I want to take a network address uh right now they find tags, 5254 and 261 and essentially have no guidance which one they should use. But in that case it's pretty obvious that 5254 are the preferred ones because they are simply shorter, but in other cases it may not be so clear. So how does the implementer navigate this?

A

We could put more information into the registry now. This first of all requires a document that that describes what information we put in the registry, but then in particular it will require continuous working group actions to actually update uh the the registry. So that's uh something we should not do lightly. So if we really want to do this, it will require some some continuous attention and there's also a little marketing aspect here.

A

We really should encourage people to register something and and not have them register something and then come in as the head teacher and say uh that that gets an f and we have a better tank here that uh you should have been thinking of. uh That's really what we don't want to tell uh people so just having a naked entry in the registry not recommended or something um that's, not so nice.

A

um So the the approach that I had in mind was to uh create a separate, informative document and the first uh attempt at that actually exists as staff from an seaboard notably takes, and this of course needs a lot of love. I mean we are back to this thing, except that we don't have to do it.

A

uh We we can update this on our own uh schedule, and the intention is to run this as a living internet draft uh for a while um and maybe actually publish this at some point when we think we have a first set of stabilized uh commentary on the tags that that we actually can try and say that is the state as of 2025 and- and we want to publish this so the only thing we could improve at some point is to actually get the registry to reference this document, and that, of course, would require some status.

A

um I don't know how how flexible ayanna is in adding uh information. uh Maybe we need to publish something to to make sure we get this reference uh in there, so that that's my view of the situation. I think it was good that brian carpenter brought this up, uh but I think we are mostly in good shape here, except uh somebody would need to find more time to actually work on this notable tax uh document.

A

So are there any comments on on this view of where we.

A

Are so christian, are you.

A

A

Okay, he's not back.

A

I think we don't have any other chair. um Can you chair from the jabber.

B

Let me try this um I I was, I was one very, very short of just um doing old, chia stuff from um via text like in the good old days. Probably um but yes, um one one. One rough comment only on this is that um I've, when I saw the slides I've looked over the the charter and I think think that we are within the charter's uh range of maintaining the sibo ecosystem, with um both both with changes to the registry and with with um providing commentary here. So.

B

um On on the procedural side, I've already taken this up as an action item for me to get in contact with people at iana to see how much, basically, how how what what it takes to just get a reference into, possibly even a continuously updated internet draft into their.

B

um Probably I'm just checking the so um probably one thing that will need to happen, and that might be a good out. Good discussion outcome for here is to consider whether whether notable text is at least within from from the scope. It has not necessarily from the content. I read it to be adopted, because then this would be easier to justify as a living document that, to some extent represent represents what is happening in the working group and then thus more easily can be referenced from the.

B

A

Yeah, so basically, we would need to evolve notable tags far enough that it can become a working group document because we probably don't want to point to individual documents from the registry uh and then we should actually uh get it in there. But it's good to know whether we actually can do.

A

B

um On the on the point of a working of um describing um older tags as not recommended, um I'm not sure what the procedure is there, but um document tags registered with from from outside itf usually have basically that entity the reg the entity that registered it as change controller um is, can can can would they be in a position, at least theoretically, I'm not sure, even from um from a normal point of view, to withdraw their registrations or change their registrations, in that they say that they are not active, actively used anymore or deprecate them themselves.

B

If we get into contact with.

A

Them yeah generally, um that's a good question, but generally, I think that that's not what I would expect to be the the standard uh procedure because well, the the person that did the registration at the organization may have moved on to a different place and so that it may just be difficult to to get the random fcfs registration amended.

A

um So I I also I wouldn't write anything like not recommended, except if, if there is really a a bad flaw, so if the tag definition creates a security problem, then maybe we want to point this out, but I think most most tags that we don't want to recommend really are just suboptimal. So there's nothing wrong with using them. But it's not not in the short list of things that people should.

A

Implement but I'm wondering if everybody else is having the audio problem as well and that's why nobody says anything or it's just summer time.

B

I think that if someone says something and cannot be heard, that would would probably result in a raised hand or a chat comment. At least I hope so.

B

But we are very, very small group today I see marco michael ross.

B

uh Andre got.

B

Oh sorry, iris iris, you are in davao here as.

B

A

Okay, in which case I'd, take this home as a added motivation to actually get work done on. The notable tax document push this up a little bit higher on my agenda, because we are now getting more and more in the situation that we have multiple tags that would apply and having some common guidance.

A

um It's probably useful for people who are implementing things.

A

A

um Christian you're saying something, but I don't hear you.

B

Now, yes, I'm I'm sorry! This is um I'm I'm at our campus, I'm inside our campus at a rome, and this is occasionally flaky I've I had hoped to pick the best spot um anyway.

B

um What I was saying is that, if, if this is to be, if this is to become a document, that's rather long-lived and that would take input from or that would aggregate input over the whole registry.

B

It might be a good idea if you could find someone to co-author this with you um just just to ensure that any low that comes from it can be balanced a bit better. But I'm aware that in a small working group this might be a difficult thing.

A

Yeah, I'm certainly open to suggestions.

A

B

Why do I just have this mental image of a group of people? I'm asking asking people to step forward, and everyone takes a step back.

A

Well, they have this mental image of somebody who got hurt and is lying on the ground, and there are 50 people around that person who could apply first aid, but everybody thinks everybody else could do it.

A

A

Yeah, I think I first have to do my homework and and to get this document into better shape. But then maybe somebody will line up for helping me with it.

B

Okay, thanks for that presentation and the work in going into the document, is there any other business that we should talk about.

B

A

Mark here, relaying from christian any other business from.

A

Today, well, if you heard me, I heard none.

A

C

Okay, thanks everybody. I think we're done.

A

Thank you bye-bye. Thank you. Bye-Bye.

A

A