Internet Engineering Task Force 101, 20 Mar 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF101-NETMOD-20180320-1550

Description

NETMOD meeting session at IETF101
2018/03/20 1550

https://datatracker.ietf.org/meeting/101/proceedings/

A

You know what my ears suck I'm, not really.

B

C

Gonna try to get started. It is 350 if you're here from the last session, please leave if you're here for a net mind come on in.

C

I'll be up on jabber in a moment as usual, we are gonna. Try to use ether pad to do our joint minute, taking the URL to ether pad. Is that last one? You can find it off the tools page and you can go two minutes and click. The ether pad link.

C

Alright, let's get started so we have a new note. Well it peels. Well, it feels later in the week. It is just Tuesday so, but you should have seen this new note. Well, it applies to every meeting and discussion that happens in the this working group and all the other working groups during the week. I think it's pretty folks should be pretty familiar with the concepts. Please just make sure you disclose any IPR that you may have related to our discussions.

C

We are doing the usual video and audio when you go to the mic, it's very important to say your name. That is both for those listening, as well as those who watch a video or listen to the recording later and get our minutes correct.

C

If you didn't like that, the previous etherpad length this one also works, we'll be jumping on two. We have two sessions. You obviously made it to the first one. Thank you, the second one. There was some discussion about whether we were gonna have that at all at the the first agenda that came out said likely to be canceled. We are not cancelling it. We're actually gonna have, hopefully a very good scheme, amount discussion on the second one and that will be focused on the version that was just published within the last hour.

C

There's a zero nine out there and, more importantly, what we want to do as a working group in the long term, considering that we now have gang library, this through working group, last call and a good, full and MBA support. So that's going to be all of Wednesday this session, we're going to start out with a couple of chartered items, Chris and Martin, and then we're gonna go into a list of non chartered items. There's one item here that got just added at the last session. So there was a meet in net comp.

C

There was a presentation of a draft by Alex and it really seems more appropriate to here at least that's my opinion, and that seemed to be the consensus in the room as well, so we'll hit that one at the end and.

A

Also because it overlaps with the finite state machine draft 15% of here today,.

A

C

The there is another draft there that we had asked the that the chairs had asked the authors to come talk to the finite set state machine people, then that's good too. That's that one I don't expect to be presented. I, don't even see the author in the room actually okay session.

C

Two we've already talked about so since the last meeting we're very excited we have for RFC's the we think that having this nice batch of RFC's is a great way to show our appreciation to our outgoing ad that we actually did some work under his tenure and think, oh it's already. It's got published today, oh yeah, so thank you been one up for expediting that processing and Benoit. Thank you for all the support and guidance during your tenure.

C

So of those two listed below one of those got published, and we have another three that are submitted to the ICG for publication and are working their way through the process.

C

So Benoit I said that he's hoping this week, which is for me amazingly fast, but but quite awesome, post working working group last call but not yet submitted for publication request. There's the ACF ACL model, that's ready for write up. Basically, there were some comments that need to be addressed. They were addressed what, in the last week pretty recently and we will get the this Shepard will get to the write-up.

C

We have scheme amount on the agenda where he talked about that and we have a couple of new working group documents, both of which are on the agenda.

C

We do have a couple of documents not on the agenda and one of them is expired and I'd like to put the or we'd like to put the author on the spot and say: what do we need to get to last call so.

D

Apologies I've just been short to time there in the aid stuff, so now that that's out of the way of the drive down the way ie. The my expectation here is: there's not that much more to doing these drafts. It's really just adding some examples through and then I think that they should be ready for working with last call. So there's not much further to do. I think the one exception here on the interface, so interface VLAN model that one will need checked with actually that they're happy with a structure. That's on structure.

C

In order to do that check, do we need to do a formal liaison, or will you just handle it through informal? You know people working in both organizations such.

D

As price don't eat sizes, you don't want and say that first you know I'm sorry I'll.

C

Do informally, I think I think.

D

That that's the easiest thing if it needs to it formally afterwards, then so--but but I can push and directly apply to do that.

C

Okay, so you'll approach them directly and if we need a liaison afterwards you'll do that right, yeah, okay, the acoustics are a little weird in here and it's actually a little hard for us, the it's a here and it's not a volume issue. It's just the sort of almost half echo that we're getting. So if we ask you to repeat yourself, please just no. It's a little hard to hear up here. Okay,.

E

C

We have a series of liaisons and coming the first one is coming from itu-t. We think it's really this one's really notable because we're being invited to participate, so the folks here in Nevada are being invited to participate in monthly conference, calls on activities that are going on there.

C

I'm not sure I have seeing this come in that that type of offer come in this way before we certainly have had very good cooperation with itu-t in the past that usually that's through our liaison setting things up, rather than just coming in the people setting things up rather than just coming in through liaison. So I think this is noteworthy, and we certainly would like to highlight it so that anyone who is interested in this area, she should feel free to participate.

C

We have a the next ones coming in from broadband forum. It's an update on OEM, related activities. It's I, don't think, there's anything any response necessary. It is also interesting for those of you that care about this topic. I happen to be one of those, but many of you may not the the next one is one.

C

We actually discussed that the last idea- and we had some draft text coming out but I couldn't see the the actual, formal response and I think it went up sort of the it left our draft and was headed up through Benoit and I'm, not sure what happened there and what's the right next step is I said: don't.

F

Wanna I'm not clear speaking so. Actually, yes, I I've seen those leeadams were a response was required. I wanted to send like an answer to all the famous. Do s right for an MDA and I was just waiting for the last piece before the schema mounts right. So hopefully it will be done pretty soon and then we've got like that.

F

The bunch of thing we want to say: okay, we have provided a store, it's published. We have all the gang module. Data that are MDA compliant I could go in a Yank italic and do a query for all the traffic R&D compliant. All the RFC and I could provide this, but I was missing. The last scheme amount piece so that the status I guess that by Wednesday evening it's not there not for any longer, but.

C

Well, on Wednesday all right, I'm, just gonna steal some thunder there we're gonna adapt to do another last call in the working group, so we won't be able to say it's all completely done on Wednesday. We just know where we're going so the note I took down is that the response was gated by scheme I'm out. We need an updated response and once we're ready, we'll send it. Okay, thank you, and this I Triple E thing is so very similar and I'd make sense. It's consistent with what Ben Alonso said. Okay with that.

C

G

H

H

Hi so I'm Christian, Huff's and I'm, presenting on the module tags drafted to Street I think that's a date. We just adopted this as a working group item.

C

Anywhere, it's GSB.

F

I

C

All right just tell me that next.

H

H

So basically, it was adopted by the working group. I just reached the same text that was adopted, but we had a bunch of changes that were indicated by the group that we should do and so I put our vision through. Also that's the O one religion, so one of the changes was to use configuration only previously, we had sort of calf moved that direction, but we had a reset tag. Rpc- and this is the idea, was that you can delete.

H

You can delete tags that designers or implementers might have put on a module that a user didn't want, didn't agree with that functionality is still there, but we have a different way of masking those tags. So we do that through configuration now, I also removed the revision from the list. Keys. I may be a little discussion on the list on this I don't know, is it possible to optional keys, I, don't think so right, so yeah, that's where that comes into play, and we also removed the King library augment.

H

So this is the old module you can see. We had a list of tags, a list of modules followed by a list of tags for each module, and then this reset tags are PC. That's like the new one is simpler. We have a list of modules indicated by name for each module with a list of tags, and then we also have a list of masks AG. So this effectively is a no. So if somebody put on like AI, this module supports super duper feature a and the user says. No, it's both bugs. It doesn't support.

H

Super duper feature a so I'm gonna mask that tag. So that's what that is for.

H

So I thought I've got a list up here. One of the questions that came up on the list during the adoption call was what is this for I I mean I, think to me it's kind of self obvious, but that doesn't mean it's self obvious to everybody else.

H

One of the reasons that we came up with the idea was surrounding the device model that we were doing with the routing yang design team, and that was- and that was basically a model that was trying to organize all the modules under this tree. So you had like a tree with the route of routing and then under rowdy. You had control playing, and then you had OSPF nice.I as bgp right, so it nobody liked that and everybody wanted the more flat model, the flat design that that mod affords less coupling right.

H

But but people do like to organize things, that's why it made sense to some people right, so we bring tags in and now we can say well, these are routing things they're protocols and then you can see the difference in coloring if you're not colorblind. The bottom tag is IGP. The for is, is for bgp, it's protocol routing, an EBE, EGP and then the third one is not comp, which is a protocol, but not routing, and it's just a man right. So it's a way to organize and counter Rhys things next slide.

H

So the open issues we do have a list of tags which are basically surrounded. You know we came up with that list based on the device model idea there might be other it would be good to if we could go over those. As a group I know, Martin had some ideas about who he wanted to use this. It's not just the device model like you could say these are open, can fix, you know, models or modules right or not, so they can they conform to the open config scheme or whatever we could use more examples.

H

Perhaps people thought they we needed and.

J

H

Say clarify the problem. You know, I kind of was like well. Humans like to categorize groups of stuff right, I came up another example: that's not module related, but a analogy would be a breakfast menu right.

H

I mean you have a bunch of recipes that you're looking at and we like to organize them too, or a menu like breakfast and lunch and dinner, and then maybe land and see like I, don't know I'm just trying to come up with examples of grouping things by tags, so yeah I think that's it, but that we don't have a lot to do here. I think it's gonna be pretty quick.

D

Local San Cisco could you go back to the slide, the new yeah model, so we had some discussion about this new structure. I thought it might be worth bringing up here.

K

D

Of the implications is that mask tags is gonna, be a list of things you're, removing.

K

D

The operational states, so, in the operational view of the data you'd only see the tags list, module tags tag, you wouldn't see the master type of UAC the tags are in effect, so this is.

K

An interesting case where you've.

D

Got some new configuring that then it doesn't appear in operational state, it doesn't appear in the apply convict this effectively. I can and negative complicated. So I think goodness to think that through and check that that's the right switch I think he is but just erase it yeah.

H

I think I when I talked to Robert earlier it's a kind of we're kind of paving new ground. Here. I I mean Cisco's sort of famous for the no concept like no interface and no shutdown right, but we don't really have a way to do that. You know through positive config, and this is the way I did it, but yeah. We should look at it, make sure it's the right way.

F

Don't want close so Chris. Are you going to use this yourself and when you have your to change depending on that I.

H

Personally, don't do our module organization, so I can't say exactly how they're gonna use it but I've, presumably the device model way. First, probably our three.

F

Ways with Vienna.

H

There is vendor, and there is yeah I- would be the local person in this chain. If you ask me personally or my.

F

Mom, what do you expect the vendor within there? So.

H

The vendor yeah the vendor, is it I, don't have expectations or anything? It's it's like hashtags right it I I picture it more as something that's it's a not an obvious thing that we do. We we organize things and people may come up with new ways to use it. The example that I came up with first or that caused it to be birth, was that the device model that would mostly be a design time thing because most you would say, I'm writing a routing protocol.

H

Ios be off right, so I'm gonna tag it with protocol routing IGP. So that's a design time now. The vendor I I'm, not sure how the vendor tagging would happen. Maybe they could be tagged with opens and bag if there I it was outside of the thing now that not really and then the user could be tagging it and, however, they wanted to organize their modules. The the vendor, one is I, think that at least probably the least one that would be used, but yeah alright, but we just left it open for white.

F

I'm trying to find.

H

There you know an app.

F

For this right, because you know if this is design time as opposed to you know it could be the lifetime, then you know the tact have to be perfect right and we've been trying those type of tax in the past of discussing in the context, for example, of energy management. Oh but I want you know if this is like or something that we're waking up a morning and then one of they combine the two. Then it's a new time, so you know, but ok I will be thinking about no.

H

I mean so I I tend to not bring up the really cool example that I I think would be really cool right, which would be to.

J

H

Api's right like if you have a comp, if you have a common path that gives you common functionality in a module, you could tag it as supporting that it's not a feature. It is a feature right, but it would be cross module and you could say that it supports that that path to that function and then you could tag any module that supported that. But this is kind of I think outside of the standards: purview, I, don't I, don't think it's as useful and IETF is.

H

It would be in maybe models that you designed you know for your own use somewhat.

L

Indifferent, so maybe I can shortly explain the problem. I have and I want to solve this, and that is when we talk to device from from an Orchestrator and the device advertisers. Are you sell modules and there are some overlap.

A

Of the utility.

L

Box, for example, there might be a native way of doing stuff, high def standard way of doing stuff and open coughing way of doing stuff, and these models overlap or they provide the same functionality, but your different modules. In that case, it would be very useful to find out which.

K

L

Belong to itself so that we can make it one of them.

L

So that's something I.

K

Would like to see I wouldn't even like to see.

L

That us and I'll stop it tags.

L

There was another comment on this about these tags. So now I have this: I have a registry. If you consider using I, know.

H

I

H

Mean the idea was to make it so it could organically be used organically.

J

H

That there is a I think the user could add them easily to the system, and so that would be done sort of at yeah, I, install routers or hardware in my system and I want to add things. I don't want to have to go and change. My.

M

Microphone Holly and racist Nachman nothing today, detail and youthful walk and I just ask.

K

Do you want to define some.

M

Metadata also similar.

K

M

Says you can, if you run into a config another operation, so you can retry this as multi-line post-attack, because I think this is only operation d'etre and if you I mean, if you try this volume, you're learning together can take two resources: module which cannot receives its tax information. Those think is big need to define some metadata as.

H

Individual souls of tap I'm, sorry, it's! It is hard to understand when it when it's echoing I, think what you're asking is well I, caught metadata right and I would blue like so point out. This is this: is metadata, wait a minute sort of data that we're attaching come on boys?

H

Can you give a clear of a shorter an example of what, beyond metadata using okay.

B

M

A nice if he lied.

M

Yes, why? Yes, why the XML tree as like this, the fuel drives, are as young technology. If you can, you feel me, I mean if you ran it again compared to the tribes is, has an etymology. It may be useful to define a some metadata, so it.

K

Can appear here, it is.

M

Necessary as I have protocol to write races.

K

Routing protocol, it's.

M

Wrote a product for ITP is this, okay is useful. Do you want to define it in such a tomography? We.

H

I mean you can define whatever you want. That's how we wrote this right. So if you're asking, if we're going to add certain tags and yeah I mean we're I'm open to hearing any sort, types of tags that you might want to add. But again, if the document is structured, so the designers, vendors or people with the devices or servers can add the tags. It's it's allowed at all three stages: okay,.

M

So it's not. We relationship it to kind of like another. The operation right, I.

N

D

I think was your question that if you do a get request, then you could get my closest future that you do get requests. You'd get annotations of the data feedback as to what the tags that your question, which I I'm not sure that's possible.

H

I know that there were little over your shooting noise.

D

A lot of extra data coming back there.

H

Are other mechanisms to actually tag instance? Data right? This is this- is working on the schema level.

H

Okay, thank you.

C

Martin also I'd like to ask everyone: please do doing either patty, but if you just add a sentence during the whole session. That's still helpful. I just said the link to the list.

L

Okay, so this is a presentation of the yang data expansions document I think was adopted actually, after.

L

So, just a very, very quick recap of the problem in the solution, so the problem is, of course, that have these yang data extension statement in the rest of our sea, but it has been used for a lot more than just within the restaurant Hong Kong people are using it in in various contracts and it has a couple of limitations. So so one limitation that was noted and notification drafts was that the structure cannot be augmented.

L

The structure must define exactly one container which might be limiting in some cases, and you also have this a bit possibly bit weird dependency to to the restaurant policy.

L

So this draft defines two new extension statements, yang data, which is known as a copy of the one from from rs.12, with the exception that will lift one of some of the limitations that that I mentioned doesn't have to be a single container, and there are some other of the limitations that are listed and then the new statement that is for moment yang data, but can be used to augment one of these structures.

L

So there were, there are two open issues. They were briefly discussed on an analyst. We would like to try to close them, so one question was: should we also define the statement, something like uses young data, so they can have you structure with a lot amended at to also use this structure just by keeping. Is a group thing, so there are also some drawbacks for this order. They have group into it, but you can use so we would have two different mechanisms for defining reusable structures if we did this.

L

Another problem is that if we do this, we would do this as an extension statement tools, model able to understand it because sorry not required, understand old extensions and it's generally a bad idea to define data nodes in expansion statements, so I would propose and I think that some of us Walker's proposed to to not do this. We just use whatever we have and use normal grooving. They use this. For, for this.

A

Can Watson as the contributor I think if you want to use groupings that the marshal designer has to anticipate that the grouping would have to be fine because they meant they may not actually have grouping to English so, but if they did that, that's great, if they don't do that, could scheme amount be used.

L

You know possibly and.

A

I think the motivating.

L

Useful case, for the other hand, I mean so this young data extension. It defines the data structure that is not part of any data store. Anything it's something outside of data source, so schema mount mount stuff data source. So I wouldn't think that you could actually schema mouth, something that is defined in this extension. That would be I. Think.

A

The driving use case for this is that you have a yang data structure, that's been defined and then that even you have yet another language structure that wants to somehow make use of it to put it inside composition of non config or operational state data structures.

D

We're losing Cisco aren't Michael either Affairs broke. Your professor fear just keep it simple. I'd rather not have multiple.

K

Ways to incentive.

D

You avoid it at.

K

Least initially, it may.

D

Be ship originally without it rope arm, comes up or definite use case. Then it we added it later.

A

Captain perhaps a way that we could address this would be in this draft say that anyone who wants to create a game to use this Yankee rat extension that they must create a grouping and then use the grouping and they're, thereby ensuring the downstream compositions can occur by using the difference. Also, oh I, wouldn't.

L

Maybe I wouldn't wonder if you require the usage of this.

C

Jurgen has a comment: I prefer that extensions do not define in a nose that's via jabber.

L

Right, the second open issue is about the error info thing. We have seen also in the notification drafts that the young data structure is is used to define error infrastructures. So error info in the RPC error is even it's an open structure. You can, you can put any structure in there, but there is really no way of defining what that structure would look like.

L

So one proposal on list was to maybe add- or maybe use this young leader statement to define the error info structure just like they do in a draft, but then also add another extension into the RPC and actions that would tell you which our infrastructure can actually occur for this particular you see currently will have to do this with description statements, plain text, and this might be a good idea, but, as someone pointed out, it kind of expands the scope of this document. It's probably not included in what, in the original scope the issues.

L

If we do this, whatever info sure we do this for a raft tag. Should we think about this in a more general error, handling document? Maybe and the last thing that can we assume that all protocols that would actually implement such an obviously also hat.

L

So personally, I would also suggest that would wait with this. Try to finish this document as it is and then possibly come back to this issue. Other comments.

G

An example just having actually living through this issue, I.

K

Just want to mention this.

G

Is important to be solved so I, don't know where the right place is always if it's not here somewhere else, sure.

K

But I think that is really a key.

G

Something that it's missing action framework and we need this yeah.

L

I think probably this is the right place right. The question is: should we do it in this document and delay? It, though, should we publish this document and maybe what of this separate documents.

O

Japan can Cisco other someone who would have benefited from this, and some of my.

K

O

Had it been available earlier, I would say, ratify this and use this to focus on something more error focused in a different document. Okay,.

A

Can't this jerk, if we don't do it now, then we would have to spin up another effort almost immediately because there are dependencies from the young push and really the drafts need.

L

To have this something I mean so do this. That would be the error in formatting I. Think it's okay to do it as they do today. Notification draft that isn't text right I mean this would be a form of way of doing what they are doing and describing in plain text. I. Think.

K

L

This we can I be more important to publish the the basis draft with a young day extension survive aching. Is that and possibly the old man care data.

A

And since the others are in the room, everyone's agreeing, okay, thank you.

A

M

P

Push and related.

A

Draft authors are in the room: I just want to ensure that they were also okay with that decision to defer the making a programmatic to know how what the structure wasn't in sight. All.

L

Right so so, for the time being, if you need something like this, you would use the I ignite extension, define the structure, and then you would explain in text in the description statement, all the RPC or action that that particular area or structure is actually being used, so we'll have a form way of doing it. That's.

G

For your time being, you should keep it, as is. This is just for future, because people yeah just.

C

As a reminder, we do have people remote and we are recording, so please use the mic and and state your name. Okay. So last.

L

Slide next steps we have some smaller fixes, editorial, inconsistent terminology and so on. If there are any other comments from this meeting or or on the list, otherwise we're going to publish a new version and ask for what last call vote for them. That version.

E

A

And next up we have Joe Clark.

O

All right in Singapore, I presented on this proposal to introduce the concept of semantic versioning for modules, specifically because, even though yang won one stipulates we should not have backward incompatible changes between modules. That implies that a module must be perfect out of the gate and that's not the case. That's certainly, we've already seen examples where there's been issues between the standard and the implementation that made it hard to move forward with yang vendors are struggling at this as well.

O

In that we coming from a vendor, sometimes we have to release backward incompatible changes and in the standards bodies. If we strive for perfection we get in the way of good and we delay and delay means people will go elsewhere. Looking for the solution, so we have proposed an alternative method in my whereby we introduce a major minor dot patch, where major introduces this idea of backward incompatible, changes, minor, introduces new features, but backwards incompatibility is maintained and patches bug fixes. So that's level setting that's what we talked about in 100 and Singapore since then.

O

Besides, adding Polish to the list of authors, we've added some additional text. That kind of rounded out some of the use cases added some additional justification and also saying that you know sometimes a new module name may be required, for example, when there is an overwhelming change to the schema, or there are some orchids organizational ships. We also added some augments and admittedly, these augments, if you're reading version, oh three, they do not refer to the this version of IETF yang library. So we have to fix that.

O

But we added augmentations to be able to add this module version or the semantic version into ITF yang library.

O

We also identified kind of a terminology issue with deprecated and obsolete, especially when you look at them as an API contract. They tend to lend themselves to a may or may not type of language which, for a contract just doesn't do very well, meaning that the client it's up to the client, to kind of figure it out just fire and see what you get back. That's really not what what an API contract should be.

O

So we proposed new definitions for both deprecated and obsolete, where deprecated nodes must still work as defined, and it serves as a kind of a foreboding warning that this will go away or could go away and obsolete, says the implement. The the server must remove this node from the implementation and any client trying to dereference the node gets back specific errors identified there with the operation failed because your dereferencing, an obsolete element, hi.

A

Joe out of one question Christian from juniper Networks when you're deprecating, is there an ability to specify a time duration for like it will be? You know, potentially isolated in one year, so.

O

We in the draft we talked about a kind of a heralding of deprecation. We it's a good comment. We don't have a specific data element that we introduced for a time bound deprecation, but we talked about it in the text as something that we should pre-warn or forewarn about a deprecation. So it's a it's a good observation. We probably should ratify or formalize that a little bit more then we we mentioned import by semantic version up I.

H

Wouldn't be in favor, actually coda.

O

My requirements.

H

J

It in a description makes.

H

Sense right and putting.

J

H

So we plant obsolete e that simple at this point. That should be actually somebody could make I.

C

Mean I want to ask clarifying question, but if it's in response the person in the back can go, is it.

Q

Eris beautiful dumb question in irregular software development I Rao, rather than a deprecation date, I get the version number that is going to duplicate with IETF process. Kind of dates are hard to predict version numbers can be managed.

Q

O

Funny you bring up version numbers because that's what we're talking about here, Chris.

C

Can I ask a question so when you say you don't want a date, is that you don't want a date to be standardized or you don't want to date from the vendor telling you when they're gonna stop supporting it actually.

H

I was thinking about as a.

C

H

So that I wouldn't want to be tied to having to put a date for moving from deprecated to obsolete I mean you might want to deprecated something saying this is not the way we want to do this anymore, and then you want to ship it to see how the market reacts right. I mean you know, you don't want to have to say on. My hands. Are tied. I have to obsolete this now, because I picked that day well,.

C

If it's not codified and it's just what the vendor does, that will allow the venatori to a nice market test and if their vendors say, if their users say, there's no way, we can accept it disappearing in six months. Well, you'll change the date. So maybe it's a good thing when.

H

There's something the data is useless right, I mean if it just wraps anytime well,.

C

The date means is expect this to disappear. Unless you come talk to us or.

A

I mean it will will definitely support it up until this date, when.

H

They're after it might what we win.

A

By by making well.

C

It's a pretty common thing to see that in software releases you know.

Q

C

Look at distributions look at it. Look at your like Linux distribution. You have the last.

H

C

Our clinics, whatever rolling.

H

N

C

N

A lot of a lot of a.

C

Lot of distros have version have dates on which they'll stop supporting their version, and this, if we just allowed for that to be done for the.

H

Vendor to publish a date, it would give us that sort of equivalent I'm, just saying don't standardize, don't require a date. Well,.

A

Okay, so don't make it mandatory true yeah.

P

All right that works.

F

That's a thing right: if you're a vendor- and you know.

K

To be duplicated and.

F

K

The version you can write.

R

F

If you don't know it's going to push yourself a constraint, just don't do it right and by the way in the text, doesn't help because meet tunings to be able to extract and get something out of it. Yeah great I.

C

O

Could I ask.

C

A clarifying question in our previous slide on.

O

The this line, or this line as I, moved on in the one pack- uh okay, it might be the next law. I know this slide. I'm presented that slide. It's one of these two I'll.

C

Wait to the end Oh cuz, then we'll see we're retired.

O

So it's good. The conversation is great. Let's bring some of that to the list as well. The next thing we want to mention is import by semantic version. We talked about this a little bit in Singapore. We, admittedly import by revision, is seldomly used, probably ineffective. It's way too specific. You have to say, oh this version or this revision of the module, and only this revision of the module.

O

Something like this import by semantic version, if done right, would give us more granularity to allow us to say import by revision, X, a major revision X and any major revision, or any minor and patch revision of X, or to import, even specifically, by X Y Z. If we wanted to- and this also allows operators to understand the interdependencies, the relationships between those modules and helps them from a tooling standpoint- understand where their tooling might break.

O

So it's it's it's important and it allows them to understand that if I move to this I'm going to also have to add or support these additional modules, so we're also saying that the mod version sub statements should be a sub statement of import and we should not include both revision and module version and, in fact, module import by revision should just be discouraged or or just killed, forgotten about it's not to say it's perfect. We want more discussion on things.

O

We have some we've identified some open issues and sounds like we're going to have a few others. um Do we need to include by some ver as well? Should we go back and do kind of a host analysis and generate a derived semantic version for existing IETF and Ino modules and and who's that whose responsibility is it to do that?

O

And what about reflecting in a module where it does change the name as to what the previous module was? So if the modules name changes, how can we reflect that?

O

This module replaces module bar, replaces module foo, for example, so the tooling can maintain that linkage because, as benoit just pointed out on the on the previous comment, we have to keep the tooling consistent simply doing this in text doesn't help tooling, we out of scope, feature and release bundling, and we also mentioned, as we mentioned last time, do we want to change the file naming notation to recognize semantic version.

S

Jason Stern here just to comment on the third bullet about, replaces module I kind.

K

Of like that idea and I can see.

S

It for simple one-to-one replacements. We have to think about whether we want to bother with cases over modules are split. Then I, don't know if this that would quite work, you'd have two modules or replacing one all going this they won't. This is replacing part of so manage. Consider one the two splits and who did you split somehow that they are material for one to one? I like this concept so tooling can see. Mm-Hmm can follow the link between the modules but up to think about non one-to-one, splits, yeah.

O

How do we that maybe it's it you have to group them or something to show that the both of these modules require to replace.

H

So my comment was on the some some of our I think um I. Think some of the stuff you're looking at is useful. What I don't understand, I think the seven.

J

Bird thing I don't see why we need it because.

H

We don't write using the patch at all right so really.

E

Major minor I'm.

H

Sure you've heard this comment before, but I mean the module name is the major. If you just standardized the actual end of the text, we like safe : version number at the end of the module name. Then you have a way of tracking the replaces. You know the the.

P

Relationship between the.

H

Major major versions right because it's stripped the : number off and you have the base module and then the version number comes.

B

H

And then revisions are your minor.

C

Blueburger I really like that approach, because it'll also allows it addresses the issue of a server meaning to support older and newer clients at the same time. So it gives you a nice transition timeline exactly.

O

So the changing the module name, it's.

C

Not really changing the module name, it's that we're now at the finding a new semantics to the month to the past the xpath. It's a note.

N

C

Encoding your major version there, that's just think that, as an encoding of the movie, it happens to be a change of the path, but really it's just an encoding of the major version. Okay,.

O

But it doesn't necessarily take into account so the import by and being able to be granular in terms of or I shouldn't, say, granular should say broader in terms of what I want to how I want to endure the combined modules or say these are the modules that I require and I can take. This.

O

Breadth of version yeah.

H

Well, I mean I, think you're addressing the case where you have a mate you actually have to in your proposal. Now you have to your versions right. You have really.

J

M

H

Is a module named Jonas.

M

H

Just backwards-compatible only major change, which is your summer major I, don't think that's the distinction here already right, I think a backwards compatible incompatible backwards. Incompatible change is a backwards. Incompatible change I don't get well. This is a really big.

D

K

To justice week back to.

T

D

One of the cases we came up for where a patch level is useful, is like a rafter on published yet module where you might be modifying example, description statements there, but it's useful to say it has changed, but the change is sort of in significance be still actually notified of the change there. The last that's one.

K

Use case and I.

D

K

Just go back onto your previous slide, prostration.

D

K

Of common here that doesn't.

D

This effectively require a new version of yang and your revision of yak to achieve this I like I like this idea of doing somatic virtually, but it gets as a question to remove tomorrow. This would we require new version of yang.

J

U

I

U

I

L

Three commands to the previous.

U

Discussions so I'm actually, first one is about that include by Sandero, or we call that so it's on the next fight, probably that's what I would.

K

Suggest not to introduce.

U

This because I believe.

K

Fell sub modules have.

U

Their share of problems and I believe if they are used, then they should be really tightly coordinated and the revisions can be included in yang, very, very information, so I believe this is not necessary to do it. It'll only complicate this thing, so that's wrong. Second, is actually two responses to to create what Chris said. I believe.

K

U

Hatch number may be useful, sometimes earlier today we discussed some way how to how to implement errata.

I

U

Are published or for analogy, for example,.

I

U

K

U

In the new revision of a module, so I think if you have a patch patch part of the semantic version, then it might be possible to keep the RFC unchanged and only if it's really.

K

U

Just to change this patch level and update the module, for example, at IANA repository and keep the RFC as it is, that might be one possibly lose. And then, regarding the idea of encoding, the major number in the module.

K

Name, one problem.

U

With this is that it will change, it will change, also the the namespace but use in the beginning. There was some discussion that perhaps we might also include include the revision revision string in the namespace URI, which was possible, but the decision was not to loose also that the main space your eyes stay as stable as possible solid. This would change this initial decision. I believe so. I would actually prefer your solution to have just the normal standard.

U

O

To your first point about, there is a companion or another graft on using a semantic version. Development type process where we feel the patch version is, is very useful in for things like those types of micro changes. A lot of.

H

Good reasons I think maybe okay, but the major to the namespace non-speaking as an operator right I want.

J

H

Of change it's backwards, it's not compatible. That's the whole point of changing the name right. It's not compatible. I can't, deploy and run a client against something where there was a backwards incompatible thing.

U

Other than actually I was one who.

H

Proposed to encode.

U

Religious derivation, not a date in in the name service. You are right, but there are other people that are strongly again Cyprus.

A

Can come to Mike.

H

Great I wouldn't want provision in this namespace because it's not changing because it's compatible right, so I can still control that device. Even though it's a newer version, I don't want it. I don't want the pleasers to change them, but for a major change, I can no longer use my client controller. So I wanted to change light, but only the major.

U

About the idea was maybe to keep your a stable, even if the changes are our major. So but you are you, are you changes the rules for updates.

U

A

I actually should've cued up this presentation by saying that there is actually another presentation following this which isn't yet another proposal, and so Krista is actually a third proposal. If you don't actually have a presentation for he's, just sort of took it to the mic. But so we are gonna have from a bake-off. If you will, after Joe's and Mike's presentations or.

C

At least the discussion not clear to me that they couldn't be used together, oh yeah, so.

G

Just to remind you that already we allow incompatible changes without name changing or.

K

G

And changing the name is rather big and people like to see their connection so I like this.

L

Movement I think.

G

You're, making pretty.

L

Fundamental changes to the young man, language.

B

L

If were to do all these studies, suppose.

L

Language that said, I think you some.

S

L

That he brought up thought up like the definition of deprecated, didn't even how to take that. That's.

A

The kind of getting further and further.

L

O

Just to sum up, Mars Commons, he says yes, there's probably need for a new yang version, but he thinks the change for deprecated obsolete or something that should definitely be pursued.

I

I think two problems, one a name.

K

I

Comp edge, but two two issues: one is that were in the process now of producing a second version of north lot of modules by virtue of Edmond da. So it could have this couple of years ago. We'd have used for them to tell NBA models. We film endgame models. So I said here, since this is a year or two too late, but more fundamentally.

K

I

Do use my software partner, Michael Kay from many.

K

Different manufacturers.

I

All of whom use version numbers some use, three-tier acid, proposing.

K

I

They have no idea where the version means into our Arsenal in fact sure what they mean by, because every manufacturer has a different sense and usually I find they're tight marketing. But when a manufacturer once make a big push once say this is something new Devon. You must have it. You have the version number it wants to run hide in pretend this is not increasing incompatible. It's causing problems in just tweaks, the patch level, only some minor number. So to me, it's a marketing tool. It's been much abused.

I

If we're going to do anything with it, we really have to nail down very precisely what we mean by a change of version, the changes level like NMDA as a change of version. Nothing else counts unless we change yang here and.

O

This graph does be specific in what these version components mean. Yeah, specifically to your point.

I

It has to be nailed down in an objective manner that we all are going to sign up to and agree with and implement for the next. However, many years.

C

As a reminder, we do also have extensions and they do have constraints, but, depending on what solution we come up with, an extension might be equally applicable.

O

So the final point I know: I've read the the draft that's being presented. Next, there's clearly thought that this is an important area for discussion. Had a slew of people come up and comment on it, people do want to work on this.

C

Microphone please.

L

O

I mean everything, I mean semantic version, handling backward incompatible, changes and being able to reflect those and their.

K

O

Are effect on other modules.

H

Ok, can we ask the question about who thinks working on a sub portion of what he talked about? Chris.

C

Just keep in mind the presenter is asking these the chairs or not, and the chairs have said that we're gonna have a discussion after the next presentation, so he's welcome to ask whatever he wants. Ok,.

E

If I can you know, keep in mind good, not.

F

So good speaking of the last time so in the ups area.

K

F

Little bit of yang State of the Union and what I've.

K

Been observing nothing.

F

In gates yet, but in different SDS, with the last instance being the hackathon with the penalty waiting like changes to their yang modules that were supposed to be standard and with the new editors Adon working on. So what I've been seeing different SDOs open-source project and also vendors, that we need to change the way we've been working right. We could say we have a good metric, because the number of languages producing RFC are going up well, know will optimize this, and if people leave that you have the perfect leg not use from now on.

U

Because when an RFC.

F

This is just wrong right.

E

Mahesh, it and LUN I can't believe that I'm you and suggesting this at this time, because I'm sure it's gonna come back and bite me on it. But is there anything you need in the draft that actually talks about how job authors are supposed to keep track of? The major minor changes is that a recommendation on is this being done in that github, just physical somebody going and changing some.

K

O

E

Then somehow they being tracked so.

O

There are, there is, as I mentioned, they a I hate to say companion, but there is another draft about using semantic version from a development standpoint. In fact, there was talk in in Singapore about running a pilot. The fact I thought you are part of that. So yes and we'll come back to bite you and in this draft in terms of how one determines what is say, backwards-compatible the concept of this derived semantic version. We've been using in a tooling standpoint, algorithmically to determine win the major win, the minor when the patch version would increment.

O

So it's not just haphazard, there's a formula for being able to determine that.

C

So Juergen has a comment from Jabbar I agree with Martin that changing the versioning has far-reaching impact and requires a new version of yak and I'm, not quite sure if that came in before after my comment about extension, so, okay.

A

And I can't I think this is a clarify. Just said earlier, that minor was a form of breaking back compatibility and major breaking backwards. Not what you have in your draft right.

V

A

Actually, backwards compatible.

V

Yes, my nose back has only.

A

When you break that competitive, is that when you go to a major, this is what's in the draft he's.

C

Saying you didn't understand all.

A

I didn't understand no.

C

S

Jason Stern I think what you're saying is. We have two forms of major breaking changing.

K

E

Module name and.

S

Changing a major version.

A

Right, if that was his other proposal, so that was an ultimate proposal, but well.

N

Maybe I misunderstood.

O

So, as the chairs pointed out, there is going to be a additional proposals and and and talk about this, but it sounds like in general, the topic of a okay, the topics of deprecated and obsolete, and how do we handle versioning and and the known changes that we're going to face with yang modules? This is something that the working group wants to work on or wants to broach is a topic.

A

It's okay: you can raise your hand if you like.

E

A

There's a fair number yeah.

M

Hello, Robin, hello, everyone. This is Michael from Holley and today, I want to introduce young with model phones, a bottle record management.

M

As way now, and the a great defender said how from you at each store- and this led a also required Morgan to compile, which is architecture that requires many module, you to be updated. All restructured, especially for some module, is already be published.

M

So they believe this MD is really good work and it can really improve efficiency, but it really brings some issues, focus on how to support backwards compatible and how to indicate some very motive. Revisions update video to address these issues when proposed this topic and defined a yummy module to help the user. The tracks of modules update to reflect the area versions, updated details, and this module can work survivor.

M

Designer aid so can help the user to improve their designer or measurement efficiency, and the next several slides I will introduce Cisneros and how to use it. London module to address this problem.

M

Okay, the first sir, it's a first just Narrows is this mode. You can serves as a model designer aid, for example, if I want to define some young demoted, such as example, fo and in this module I want to import some sets of modules like I'd have interface.

M

These modules may exist multiple versions and before before the design, time I may not realize okay, this model had the update and it just her modules and leave already detector kid so maybe I can I can meet, makes a mistake to define the interface to find a define, the leaf, which is pond to some path. This path already be deprecated in the listed version of interface young module.

M

If I run in this mode use, maybe I can receive some arrows like here so I'm a confused, oh I, make him feel why I receive this error and what makes me think what mistake I have to so. If we have some tools to help me to reflects Aaron Lorenz update, it maybe can help he hadn't made to quickly fix its arrow and if I read this updated details before design hard, and maybe I also can't avoid this here. So I believe this Mon you.

M

If how a young in multitude will fly sharing reference of the digital, it can help me to improve designer efficiency. The succulence neurons is this model conserved either management aid, for example. Some a winter's sister may can support, but the version of a sim module, but the system can only implement one reverted and how to make this decision and how to do this work. It here lacks adapter, where the system can support a somatic host.

M

Your the inter, inter phase of reversion and if I can hear provided to choose to help the model system management to understand on which we versions fascinators, that it. What is a cat eye back workability and watch it change logs and either kind of help me to make this Aryan race. Religion should hire impediment and I also can serve each module with this model, make it user interface I need to be update if I synchronize I have a decentralized. It's very related modules said: I can implement this diversion.

M

All I can first implement the first version and then do seven such access updater utilized, then I can modify, choose a place to worship. Okay, the Leicester scenario is a small team. Can work died as a traitor, so Folsom water designer for thier module? There may be really needed to track the multi, which a important import of which gate and down the status if I have provides it. I can provide the tools to tell the designer each versions. Change law. I can help me to quickly synchronize the modern which I am designer at all.

M

Okay, this is what modules over real inhaler I way, provide the anti DiMaggio and as a module for frustrated, provide some basic information such as the module, a introversion, and it was a backward compatible information and behind it's this even basic information when he provides a motive change lock. It present this updated detail from this world and that's a list of world and the previous version.

M

Here we define leave change operation. It can be indicate what kind of update appear in the list version. For example, if I we have deleted some statement, if we have moved some it no to some new location or just away a de niro cement or modify some statement, and also here is to case one case for some definition modified on the data definition such as a stiff leave, definition least, container and Troy's case rated hill of modified ounces nose.

M

It can be easily present in this module and another is another case, a the for some update which, located in the I, didn't a definition, feature definition. Tab definition such as grouping definition since also need to be reflect in this module service, and we also defined RP State to help the user to compare an a version of this same module to tell them or what kind of update they have to in this version and another one previous version here is some uses. Example I.

A

Can't watch the nitrogen super as a contributor. Why are you defining a top of a model instead of augmenting game library? Did he think about maybe augmenting library.

M

A

M

To yes, I sorry, I forgot you provided to stage says this model, also augmented a young elaborate. Is they define the Sun live to reflect back work happy dating sorry for that? Okay, so, here is the same use case is prison how to use it module to return to the update. Half I'd have interface module.

M

The first graph is how to use motor present delayed state sub tree, and the second is how to age new feature in another definition case, the third graduated to introduce how to use a cyst mounting to present how to add a new sub statement like a feature.

M

The last is how to move some loose and didn't note to a new location, yeah.

M

The next type of the lock to so they say the more commons and making prepares another version, basically comenzar feedback. We can go out document and upper demise. Our younger devoted and killer question.

D

Robots in cisco, so I wonder this is solving a different problem than the one in the previous presentation, but I'm not sure I'm. The.

W

I see how to solve sort.

D

Of version issue, it's more like there and yank out a lot of extension type idea of providing more information about what changes have happened to, but no I, don't I, don't see how to solve the version at all, and then one other portent. This is is whether this is something that's useful to standardize or whether this is just something that we provide. Advice and tolling did the dick between two gang modules and just said that at least the to these.

D

The changes between these these revisions, which I think is useful as useful information to have whether hauling in solution, would be better than a standardization approach. Maybe I'm.

M

Not really sure whether it is better to define this realization how just difference and chooser, but why we think, if you can hear it is information of such documenting, the American helps user to track the modules, teachers and track the modules updated.

K

M

Whether is standardization or whether, if a tooth, maybe we can a more discussion and that the working.

I

Group can exist.

N

O

Joe Clark Cisco um I agree with Rob. um You have a boolean node in there backward-compatible, but after compatible with what you kind of have to say, so we built as part of gang catalog org exactly almost exactly. What's in your draft meaning and in fact, in your example in your draft, you actually have a source and a target semantic version 1.0.0 compared.

K

O

We build the tool, I need to do that and what we found is semantic version doesn't standalone by itself. You have to compare it to something so having just a simple boolean for a module that says, is it backward compatible, doesn't have as much information and saying given these two versions of this of a module with this name? Is it backward compatible?

O

Is the newer one backwards and either the older one and what are those chains and, and so I think from the tooling standpoint, there's huge value in this we've already seen consumers like that meaning they appreciate the output of that. But I don't know that alone. What your solve I, don't.

K

Know alone, that the.

O

To our proposal and your proposal, conflict and and I don't know that that there's anything added that isn't solving tool. Yes,.

M

I have written your document and I think a theoretical to walk and I. Do this thing as Eric, and you know I, don't think to have some overlap. I believe that to solution can compatible with each other, and the user ocean can choose user. How changing is changing illustrators status? What is a big change? What is a mini change or pad change and the our solution? We want to provide what it's a big change of patch and really are yeah I think they can work together.

A

The enter both parts I, so yes, they're compatible, but what about the tooling there is. Is that what's the value added above and beyond, which one can do.

O

I was agreeing rather than that point that I think there's value here, but I see that I don't know that the standardization is needed. The tooling is absolutely needed. This output is absolutely needed because module consumers do appreciate it I, just a I, don't think that what I presented him, what was presented here, conflict- and he I don't know like Rob- said that it needs to be part of the server itself versus some external tool.

J

Carry Nokia so as a as a module provider right I would assume that the expectation here is that you would expect me to have to develop these artifacts right of these particular modules of saying, I move this node from here over to there this node and increased its these increased its you know range from X to Y and put all these together and some sort of change log of, if you will, in order to be consumed by some extra.

K

J

You actually expect me to do that. That's the problem. This requires two things for this tolling to work. One is for people to provide it, and the other is to have the the ability to consume it. I would suspect that, if we're having problems with vendors already that.

K

J

Say I'm not even adhere to a major minor patch. You know guideline that they're gonna be able to. You know, build all these individual constructs and archives. So it's a great idea I'm just wondering if, if indeed the industry would be supportive of that as they define their modules and have that particular rigor in place or require that particular rigor and that's that's my biggest concern with something like this I've been through this.

J

You know down this rabbit hole once before, and you know you're not going to come out on the dark on the bright side of it.

A

And maybe are you.

A

O

I guess any more questions address that so Joe Clark again from a tooling standpoint, what we've done we.

I

O

Any vendor to produce these artifacts, what we've done in our tooling Elise is note. The differences, use P, n PI N and show you a report of buying and then show you a contextual diff of the module. So we don't expect any additional tooling or just showing you what what tooling is available, what those differences are not to.

K

This degree, but.

O

At least you get an idea of what change-o.

S

Kind of on a similar line, I guess I'm wondering who's gonna consume this detailed information. I can see that being useful for a programmer of a client. In order to say oh I'm gonna support a new version of the module. Let me see what all the dips are and I can adjust. My my client.

K

S

K

Kind of a that's.

S

A that's a upfront, implementation time, information I'm, not sure whether a client would dynamically want to query this list, and then is it really gonna be able to use that list in some way that's different than just looking at the new yang will.

K

S

This leaf is obsolete. This leaf is deprecated, I'm, not sure what what the clients gonna do with this list of dynamic information inquiries from a node hard to imagine what it can do with it.

A

Can't as a contributor I was actually thinking the same thing. That seems like programmatic release, notes and, and you know, but already you, if you have this as a standard and there's an RFC and then there's a BIST at RFC, even I have something like changes from where you know, there's gonna be a section is describing the changes were made for the last RFC, which you'd then capture some of these kinds of details, the release notes. So whatever wise is needed, programmatically I, guess. The question Jason's comment.

M

Yes, maybe you know, maybe sana may have some assimilated system and I need you with. So they say to re reversion of this module of their system in Shu support every version of this sim module and then can take, can chance at least often already change at all, also to proportions and make sure ok, every related module with it can be centralized off my plate.

A

All right, thank you, Michael. Now what the discussion I I, think. We've all agreed that this is compatible with the other silver and Christmas proposal, but yet different it's solving a different problem. We probably the discussion row should just be between those two proposals: the Zimmer and and main mangling of the model name and ten minutes. I.

C

Don't see where we are after ten minutes of discussion and if we ate, we won't artificially cut cut it at ten. But if the energies dying down we'll try to wrap it up and move on to the other presentations that we have.

N

Michael I've run some soon I, don't know how to solve this. I can just assume user and operator in your safe I've had a lot of pain when it comes to going between different person like we're. Developing the I malls were all scissors. This is a free open-source project, I'm putting it on this.

K

Home gateway, some.

N

Revving the model further quickly because we're developing it and I'm in driveway with an animus that needs to have changes there. I've had so much pain in the whole development process, because we're trying to create one and a mess to run everything. So there is any kind of device coming in has to be compatible with whatever is on the NMS so and also I have different verses of my case. The whole gateway that support different versions were to try different prefix different namespaces and they're.

K

Different model names and.

N

It's just there is it I've.

B

Had a lot of pain and I don't know.

N

If this is like special property of the NMS.

K

N

Using or something but having this over time, how this works, how to work with this yeah but I think we need to be more.

A

Things goes to the importance of the problem, yes and I believe we already think it's important, in fact our ad right. Okay. So that's what.

C

I understand the next hearing.

N

B

C

I think is good I mean maybe I'm cutting pent off your sorry.

A

X

Know hearing all.

C

Views and particularly user views I think it's really good and if you have any insights on specific areas, you'd like to see address within that, that might be something that'd be helpful. Yes,.

N

I don't know we need some kind of implementation advice or like what not to do is well because, like I have had you know, the device is people testing software upgrade down drains and the device is flipping back and forth between the models it supports it. Does you don't support the old one and then you know only supports new one and and then like? How do you handle that if the yeah that's like I, don't know? If that's something we need to cover? That's obvious, but the.

J

Whole thing in like.

N

We're following the ITF processor were implementing different. As the draft is progressing, we might say: okay, we're gonna, implement it.

K

N

Know whatever is in the draft right now to two revisions later the model has changed, and then it like okay. Well, what do we do now? Eve? It should be handle this with different prefixes and different model.

U

N

Or punch with country to do that like? How can we do this as a more of a I? Don't know DevOps approach or something where we continuously changing things, because it seems to me that so far the whole yang model? The thinking has been you putting a lot of work. You drop them all, and this is enough static for a long period of time. You might add a few things to it, but it there are no major changes. This is actually not what's happening. I think the real world is a lot less serious.

H

So I I work with Michael but I want to echo I want to I, guess sound another one of my colleagues real pain points. This is not particularly an IETF thing, it could well, it relates to modules and yang. It doesn't relate to something I think the idea can change, but one really painful thing that we've had is that vendors are changing the modules of a ship like with every release in incompatible ways. I mean it's, it's really obnoxious it's so we see.

H

Basically, we have to treat every software release as a that's all of the modules Frank right now. I, one runner just doesn't change anything. So the we just get you know we just the the module, doesn't indicate that it's major changed, but it is. We have another vendor who just puts the incrementing number and namespace right.

J

H

It's like the extremes opposite, you know like so. Every single route.

Q

H

It says it's backward backwards, incompatible when in fact it might not be it might be compatible. So anyway, I you know it speaking to the vendors out there right, stop doing it's really painful and and.

J

Eventually, you.

H

Know right now it's not affecting us buying your gear, but it will really soon so.

A

I can't as a contributor and Epping juniper is at fault with this as well. Every release we essentially it's module name or you know Juno's underscore release number and it's and there's no guarantee that anything's backwards compatible because we can't but going to your other drop, the module tags draft, so there's standards, modules and then there's vendor specific modules. So maybe this would be very beneficial for standards bodies right, which is our personal pain point, and then vendors can continue to be backwards and pedal hybrid releases as they are today, but.

G

Angular Exxon I'm speaking, both as a device standard but also.

K

As a OSS vendor.

G

What are all assessed walls is that have some simple data like Sam there, where you can immediately determine. Is this backward compatible or not? Okay, anyone can go I. Am we think? But theoretically you can very fast determine if it's compacted or not, and then you can try to use your software that was prepared for the other compatible version and that's fine. Today we don't have that so their particular cases and obsolete is kills backwards-compatibility practically.

G

On the other hand, I don't want to change the module name or the name space, because all my scripts, all my software, will stand down that namespace and that module name and all very often some parts of the more you will exchange that I. Don't care about my being compatible but I?

G

Don't care about I want to know that there's a change I have to investigate it and that I can decide what to do about it, but if I am forced to change or buy scripts and software, because every time the more you learn to think it's changes. That's that.

C

Transfollow fashion on that flew when you say you want to discover, what's changed, do you want to do that with the device, or would you like to do it with offline mechanism like a catalog based mechanism? What.

G

I want is to immediately notice if there are backwards, compatible changes.

C

Or even you're, making it you're getting a version number from from, let's say the device. If.

G

I get the Stamper, that's good for me. What the changes are. You can't find out.

K

G

You lose compact were incompatible really programmatically. Very often we have the case when the model is not changed for the behavior behind the model, it's changed, so you will anyway need to study manually, though, and I would say that this detailed analysis, it's very, very useful, but I would say that. Yes, that's.

G

C

So the what Joe described would actually work for you in terms of his the card. Catalog, actually I'm,.

G

C

Girls on that document.

G

C

F

W

P

F

Uncertain to forms one of yours models, because you.

K

Know we stick about standard. There is this entire world where.

F

We've got vendor-specific models that are generated so and that's a fact right. So whenever we have great wealth, the microwave, so we could play tricks and against understanding model so that we will be respecting 916 videotape right. So, but you know it's not working like in the hops way. So you know what Joe mentioned about the catalog. What you're about to do in a catalog is you get like two iOS versions, but they need to query that because the metadata, if you're, going to cut a lot.

F

And yes, I was.

P

Mentioning some.

F

Time it takes them on your process, so we'll show you the tips of the three and the module, so you know yeah. It should be working offline. Now. The second.

B

Thing I wanted to answer.

S

F

Your fun about the feeling, the pain, yes, I- think that we need to change a lot of things here. Let me try to be both I would even try to still people must be changing the process here, and it's easy for me to stay since I. Want me the ID like that more Evelyn, but you know we need to be DevOps, focused up here and be focusing on those young models. What we did, because we did all the experiments.

K

F

We thought about in the young catalog in there.

K

V

You, like the house metric.

F

Of a young module right, so whenever you go and you tell us, every single draft I go and implement it and it changes things my script, etcetera. So the process should be change in a way that you know, we've got those health measure exact. It validates it works with the other yang modules. There is code, it is imported and all this, and only at the point in time, then you know that it's kind of ready. You feel free to be implemented now I'd.

F

If I want to go one step further, can we just run our idea process directly into this? This yang module? It is what Joe was mentioning with the I am drought there, and maybe it is this, like we run the center on the on the draft in github right. That would be at least I hope helping you deliver a mitosis, but that's a big change all right. That's a big change.

R

Charles I thought so a few different comments. One I agree very much with what I think Chris process. We don't want two different ways: any major change or in competitive. You don't.

K

Really care big versus.

R

Small that, surely, is it non backward compatible or is it backward compatible the critical thing together that and as to how to encode that, whether it's a different URL or namespace, or it's actually encoded in the model, the the.

A

R

N

With the philosophy product.

R

Made a lot of sense to me for going Joe's method is I could certainly see if an operator wanting to do that and then the last thing is that there was a comment about. You know this, maybe being too late for in some regards, but I think through the need, for this type of thing is only going to increase, seeing.

K

A lot of requests.

R

Or hey, we need to move faster when we move need to move faster, we're not going to we're going to even more need to change models, so I think. Whatever mechanism we do, we've got in order that it works very well for lots of compatible backward compatible changes that that's done to work very. A little.

K

Touch easy way to.

R

Deal with that just yeah the faster we try to put out yang models, the more we're going to have to change.

A

Okay, I want to follow up on some of the comments that were sorry. This Kent as a contributor I, want to follow up on some of the comps that were made earlier during Joe's presentation about how, if we were to try to implement it, it would require a new version of yang Yang next and, if that's true, that would be a long term solution. We really need a short-term solution. So can we work? What is what is the word short-term solution? Wow.

G

Actually, we are already doing young expansions for exactly the same reason. Why not idea works so.

C

You're saying any extension is supposed to be yank version. Yes,.

G

That's what we are doing.

C

G

C

G

Yank version America.

H

Well, okay, so I'll count that idea earlier right, I, don't think we need to change the language.

J

H

If we were to put some semantics on the URL for the major.

R

H

The revision as a minor I'm sympathetic to the people saying the patch number was useful, so maybe we need to come up with some way. You know so, with pre, with already existing mechanisms, we can have the major in code to the namespace and the minor is a revision. So then it's really just the way of capturing the patch change right to say: I, didn't change!

H

You thing with descriptions here, so don't worry right and I think that if we could do that and incompatible ways- and it would be easy to push this through right- start benefiting hey.

D

Brooke Walton I'll just make the comment that there's any some parts of the solution that Europe would require change it to.

U

D

Just adding in semantic version.

U

D

Extension or protection is fairly easy. It's if you want to have the import file and revision herb by somatic version.

K

That's the one will change the.

D

Semantics, the language will require new version.

I

Tom again hmm well, as it is too late, bearing in mind that we do have.

E

I

Revision by day and when it's made, that's not much used because to make the questions. Why not, but is we do have a former version currently based on date, if it's not being used, we need to know why it's not being used twice inadequate and that that was really driving. My thinking when saying we're too late to try and impose a formal structure of major/minor patch and so on, if we call baton, we haven't used it. We need to look at that as well.

C

So you're gonna is actually in queue. That's okay, name spate. This is from Juergen. Namespace is used by XML. Json uses module names for the namespace, so you have to change the module name for a major revision.

L

Y

Mean in level yeah.

L

Y

If you're, not in this.

L

You know major version of it to the to the module name.

L

And, of course, this is what the current a.

Y

L

For normal backwards, the particle change so we're going to comment.

K

On calling Rob.

L

K

Just said, it said that if we just.

L

Change any yong, we don't have change language, but.

K

If we had automatic version.

L

With the semantics that they proposed, I mean we still.

B

Need you to change the.

L

Language, you wouldn't.

F

Be allowed to ever.

L

L

K

C

L

C

I suspect every we have to look at the details, but I suspect you could also add Chris's as an extension and because it's because it it would be compatible. If you don't understand the extensions it just looks like a new module, Chris is, is more conventional yeah, except that I think you would want to do inclusion by name still, where you don't care about the version, which means that that changes, the parsing of that.

C

So it's a it's a little bit more I think it's an extension, not it's a little bit more than just the extent adventure makes you think to make inclusion, work.

H

I would not say that that's.

C

A pretty major change you're talking about like wild carding, the version.

H

Number I thought.

C

You were talking, this is your.

H

C

No, that's what I'm saying I'm not saying.

H

H

Q

H

Give people what they wanted, which was some some pathway or a like I, went from it's the same base module. That was one.

J

Of the people's complaints, as if they couldn't tell.

H

Like if you change the module, maybe you have no idea where it came from. So if we standardize.

J

The version at the end of it, we know right.

H

Easy to see, then you don't need a place to.

C

So, while I think we expect what the answer is here, we're gonna ask it anyway. How many think the working group should be working on the topic of handling modeling model, provisioning, I.

C

Must have asked it the wrong way. Cuz I would actually I want to hear from Martin or why his hand didn't go up, and that was.

A

Much the room yeah.

C

I expected everyone in the room and- and we got a lot of people but Martin like half raised and then went down. I, don't know if anyone saw that in the back is this? What does that mean? It may mean II didn't like the question.

O

C

Think I accept the clarification. I mean versioning.

C

Without meaning a solution by the way, I don't mean semantic versioning.

L

I, don't think that we shouldn't discuss it of course, but I'm not sure that we.

C

R

I, don't anything.

L

C

Agree with you.

A

We agree yeah, it's just that blues. Just asking. Is this an important problem to work on.

C

So I have to I have to channel from the floor and I'm gonna from the from Jabbar and I'm. Actually, gonna start with an apology, one of the things Kent and I didn't do on the first slide, because he's not here is forget to talk about Joe land as being our new chair co-chair. So Joel is our new. Is a new co-chair and he's actually online, and he has a comment so and.

A

We were going on his IETF profile picture, it's actually rather humorous and we're gonna put it on the first slide, but we didn't know.

C

A

C

It's on the tools page if anyone's interested go look at as the the tools page that mod and at Joel's picture it's quite good. So his comment is agree that it is an important problem so back to the view.

N

Then we're not good at using them, then whatever is available right now. So then we need operational guidance on to work with that. So I, just I can just say: I I had pain, I, don't have other quickly. The doctor please make.

C

Me that so we were gonna, ask I, think, there's three questions and then the next step was gonna, be working on and identifying what problem we really want to solve. So maybe I'll ask the three questions and then to talk a little bit about next steps and then open it back up to the floor. Okay, are you you really want to comment I.

G

Just want to say that, yes, our current solution is not good because of the deprecation obsolete problem, because I can't determine if it's compatible or incompatible. And yes, because of changing the name for spoon in compatible error. What we have under vision, problem, I, think I think.

A

The consensus in the Romans that it's important from the work on right.

C

And I think the the the point that was made is is we should clarify what exactly what Martin is leading to I think we should address as well what what problem are we really gonna try to solve here so I think the first step in addressing this is capturing that how many people would be willing to work on the problem and with the understanding that the first step is to define what problem we're gonna address.

C

So of those very reasonable number of hands, how many are willing to offer text.

C

Okay, we would like your names at the end of the week, we've got a reasonable number quite seriously. We'd like to your names, send it to net mod chairs. We're gonna talk a little bit Jules, not here.

C

We can't make it a definitive, correct direction statement right here, but we're gonna talk about maybe doing either an informal or formal design team, and obviously we also have to check with our ad about that, but it seems like this is important enough that this we have enough people who are willing to work it that we could get a group together to help focus the effort.

C

Martin is resisting saying something of the bike.

A

C

And now I'm done so Mike is open again.

L

C

A design team to go work, the problem and the first step will be the problems they fit and it, along with any design team, comes a design, team, Charter and deliverables, and you know so. The question is: are we going to do something formal or do these people say they're willing to work it and then they're gonna go willing to work this week? Hey we don't want to get in the way go. Do it now.

C

Any other comments from the floor all right. My apologies. Are our colleges again to draw.

C

Thick we're on finite state machine right. Yes,.

W

So this is the version two of a young model for finished state machine. So we we proposed this model with the aim of increasing the product or mobility of the network elements, and we are thinking to to apply this. This model in three use cases of applications. The first one is to pray instruct optical transponders about reconfigurations in case of physical layer changes. The second use case is related to telemetry applications.

W

The third one is related to performance monitoring through a network clustering approach. So let's go to see the first use case assume to have an active service connection. The Sdn controller can instruct the the optical transform transponders on the actions to do if something happens at the physical layer. So if a degradation happens, the transponders already know what you do again. They can promptly react so without alarming the Sdn controller and waiting for complete patience because they already know what to do more.

W

Details can be found tomorrow in the sense, in the session of the CCM. The second use case is related to to telemetry. So the aim is to customize Network data collection with the objective of reducing data data bandwidth. So in again many many network probes can be described through a finite state machine. So in this case we can program the elements to to trigger data, sending data telemetry stream only under specific conditions. The third use case is related to IP performance measurements on multi-point to multi-point large networks.

W

The case of point-to-point transmission is is included here and, as the end controller can calibrate how to how deep we can obtain the monitor the monitoring.

W

So we we can. We can think of the about the alternate marking, meter metered for performance monitoring, and we can apply this into a clustering approach. So we can image to divide the the network in in clusters. So under normal conditions, we can think about large clusters. But if something happens, we can reduce the sides of the class. That's going so more deep into into the the performance monitoring up to the flows again, a Finnish state machine. This can be modeled with a finite state machine where each state represents a specific division subdivision in clusters.

W

So this is the young model. It's you can see the current state. Then there is a list of of states to each state. There is its associated a list of transitions, for example, for the use case of optical transponder. A transition can be triggered with a bit error rate, increase the transition as a filter attribute, which permits to further describe the transition. So with these attributes we can put, for example, some thresholds on the bit error rate increase and then to the transition.

W

We associate some some actions, for example the change of modulation format or the fact for the for the use case of the transponder. The attribute execute actually performed his dis action. Then we go to a next action, if present and obviously to annex the next state, one AD when the the transition is is completed so.

W

This model was originally presented in was submitted in June I discussed. The first thing in Prague, obvious AWG, then has moved to net mode. We with this version we fixed some issues related to the syntax and we'd like to ask for for the adoption of the work of this working group.

S

So just a general question: when we trigger an action and then when you trigger sorry and then you take an action, you call it RPC, so I assume the idea is. This is going to do maybe a configuration change, just as if it was any other management client like there's no special interface, it's just as.

K

If it was a any other like net.

S

Comp client.

A

C

A

Need a microphone.

C

Please Mecca or the.

W

Use case of the transponder: yes, there is a configuration action, so we reconfigure the modulation format, the the and the fact but well. Okay, then yeah so.

S

So yes, one issue that comes to my mind, right away, is at the same time as this may be triggering. You may have a management station also doing configuration which may have a lock well.

W

Considered that once the the the its reconfigured we can send another now another action can be implemented to send a notification to the controller.

S

Sorry this this action, which is trying to do a configuration change just like any other client, it's gonna, get blocked because the management system is doing took a walk. It's doing some edit configs doing a commit that could take seconds or a minute or, however long it wants. If this is just doing a standard, commit like any other, a standard configuration change and commit like any other client, then it's gonna be stuck behind any locks by any other management stations right well,.

W

We we can, we can think about how to how to handle these vision.

S

You know in a special sense. Your use case seems to be a very time-sensitive use case. Okay,.

W

We can think about how to handle this issue in in in future. Okay, thanks.

O

Joe Clark. What so in this example- and this is what I was kind of- maybe related to the previous comment. What's to prevent so in this.

K

O

Error rate increases, and you call an RPC what, if nothing changes, what what in that? This to me feels like a infinite state sort of like a if interested.

W

O

Is your ability like edge triggering what.

W

If nothing changes that that it means that no transition is triggered and we stay in the same state in the current state.

O

This continue to execute the RPC over and over no.

W

What no, if, if the transition is triggered? Yes, we go to the action. We execute the the new configuration we go to the next action. If present, then we go to the to the next state. We are in a new state and let's see if a new transition is triggered or not, okay,.

O

I was thinking constant error rate increases no.

W

Okay, now I could well if we we set that filter attribute for for avoiding these continues. Look yeah yeah thanks.

T

Actions can be excused excluding synchronizer how in parallel, so not sure you, you are considered situation, not yeah company predict so HM maybe in depending on the activity and then should be executed in advance of action P. So there is maybe a detection mechanism that to detect this kind of company.

W

If I understand correctly, you said that about actions in in sequence or in parallel.

T

So I mean I, think I I killed it in advance of a kewpie.

W

Well, we can, we can execute actions in parallel. Sorry, in sequence, as you see, there is the first action that recalls a next action, but we can also think to execute actions in parallel. It means that we we have to define an action that intrinsically implies two actions to be executed in parallel, I'm, not sure. If I answer to to your question I answer, it did I answer it: okay,.

A

Thanks: okay, so that's net launcher. We were. We actually have the author of the other draft in the room and we're gonna offer him to present his idea as well, but at the end of the session other with last presentation as NECA co-chair, he just presented his idea in the neck half working group and there we said we felt that he should bring it to the net mod working group, no so bring just like you brought your draft from ops no later yeah.

W

A

Back to being that mod, chair or co-chair so at the end of that presentation will then see if there's an interest in the room to work on the problem.

C

P

With Igor yeah, the.

C

Next step is to talk with Igor offline or, and hopefully we'll have time for him to present, but we may not have time. Is that what we'll see but either way? The next step is to talk with them and see if you could get them work together.

A

Now we have Jim.

V

Hello, my name is hiu the topic. Our presenter here is a document document convention for another long line in autoworker containing code Shanksville shared for suggest me actually only way. The title title is actually experiencing young in XML example. By the way, single would also need to consider jason cases, so we change the title to to make a to correlate with it, so so why this job? Actually this job? The way when we actually walk on air 3sn actually get leads to UPS a few n-no one issue actually come up with.

V

Is we see in our CS and BS? Actually, that's quite a lot of example: X X X Mary example. Some online extremely known so exceeded the line. Names mimic actually usually set up cemeteries character. So it's so talking with LPS SSN didn't came. Actually we one solution. We come up with broken.

K

Line into several line, but.

V

This actually is.

K

Not allowed based on the.

V

Young 1.0 1.1 in update 7950, so also we we can reference to the obviously sixty eighty seven piece. Actually, even if you emphasize actually example, each an extremely important.

K

V

K

To be validated by.

V

K

V

Do some investigation in some existing already published the FC or some some some book image of that working in progress that we see the XML example.

K

All chasing example, actually.

V

After I talk to me in like a consistent way, so that that's why I can't we come up with the idea we want to provide.

K

The document invention.

V

To avoid these kinda inconsistent issues, but the we are here, we want to vomit I way to have no change to the rules to present a young module in a fire on a on a wire. So what does this job do? Actually, so we need to avoid each talking to provide its own maximum to document the XML example chasing example. So so we actually define several convention. The post is convenient, for example, leaf would not value across the nine.

V

The second and convention is about representing X narrow, encoding of the metadata, a notation in the Huey actually define rules to passing the.

K

V

Xml form example, and we also provide the boy played text to use when this convention are used so poster we can take a look at this example. Actually, in this example, we have done not the label and long, not a value, so so this actually actually exceeded the known, then the nine men's limited. So so what do we proposed? Actually, we can proton this leaf, not value into several lying actually.

K

V

Broken line actually must be terminated with X and without any additional space before the backslash and also without any character after this backslash for the continued continuation line, it must start with what whitespace offset at this two parameter. So also, we add a statement to indicate actually the document convention.

W

V

Define a year actually is applied to this kind of example. So the second case actually is there, is a more complicated and in this case is the way we are actually Russian.

A

V

L

B

L

Like so the second bullet, they say, any continuation line must be indented with the white cells. Of these two characters is that in the land for ways that yeah.

V

L

V

Gave the example, in example, P yeah. We can see there's some indentation with wider space, that at least two character.

L

V

L

White spaces instead of two that would be allowed, but how would the I already know it if that's one read white space or three white spaces, but it's descriptive I mean.

V

Companion contain can repeat.

L

So you say you say that it will be indented with at least two white space characters, yeah.

V

I think that means that.

L

V

Precise then, maybe we actually met here the example which has to set.

L

A to character.

L

V

K

J

K

V

L

Yeah, so the other problem is that in XML, in a leaf like this, all white space are important right. So even in your first example, you actually have a new line. First yeah.

V

L

Right so that should probably also be close or escape.

L

We never need, even if, in the top example yeah.

V

L

Start helping with a new line, oh.

V

Yeah now that's a maybe because I think in in a.

K

V

I think we already fix these naturally yeah this here yeah, it's not a consistent Samsung point addition.

U

V

So second example is, more communicates actually may include some Pakistan she in leaf nor the values. So in these cases that we said the convention is when the package appear in this league nor the value the example master arranges.

K

V

The package is not a fan of the character in book nine additional. We have some case to apply to the metadata interpretation, so here we gave the example we see. Actually the leaf, not a demon is that it and include the meditator annotation, actually a nice extremely long Zoe. So the convention.

K

We stand here is actually.

V

We focus on one nine into the several ninth so similar.

K

Convention applied to the.

V

Medicaid annotation Paris, so we downloaded the community nine months that awaits the widest space, offset here of the two characters yeah. So in addition, we different from the the previous exam. Actually, we also require the leading and change Trina training y despues actually must be Eagan on and so.

B

V

Added a statement to indicate the convention we defined here to apply to this kind of example. So next, naturally, is the rule. We define a for depends: the value the X X now example. Actually we have a full quality. Actually, the first the topology actually apply to the long leaf, will not value.

T

V

The last that you already actually apply to the metadata.

K

Invitations so.

V

You can really get hair in chapter for this, so we we also be actually this none of the biggest problem.

K

In the world we.

V

Think we need to solve the incident so for next step. We actually talk with the cars that we think we need to add the chasing example also for the rule to define the passing the young different either XML example. We should add the example: validation.

R

Automation by the.

B

Tuning so this is we.

V

Plan to do for the next step, yeah comment.

A

I can't as a contributor, so first but I think this is an important problem solve. If you have very long lines and you do a submission, ID Nets will give you an error. It's not a warning. It's an error that your line is too long and in a lot of my graphs, especially the ones that contain basically for encodings I mean like there's almost a paragraph of text, so I do believe it's an important problem solve my second column is I. Don't believe this is a unique to XML.

A

I think any artwork that contains long lines are, is subjected to needing to be trimmed or dealt with. My third comment is my solution. I'm in the acknowledgement section. Some of this came from work that I've done already. It might some I build scripts when I'm compiling my drafts, essentially it it I, have an ability to do a fold, the UNIX command fold. You know exactly place the backslash and just it lays that perfect, so can go and into the draft easily. Now it doesn't have the indenting.

A

You know along a column or to space or three space. It doesn't have any of that, but it's readable, and it's not subject to any particular coding syntax. You know XML versus a sauna, it doesn't matter and salts are very easy to reconstitute the original. So I think those are my comps.

I

This is the problem which you need to XML. We restrict myself purely to our CID is containing yeah.

R

I

We have a line to the own problem in the tree diagrams, which we discussed to some extent already. We.

K

I

In one or two modules in the bar statements well get got those slash, dot, dot, slash, dot, dot, slash and we end up getting edited characters. So we need a convention for that and then stepping outside Netcom altogether. It is in the census on problem that it pops up regularly in security area. We have time for accent, keys in matters, you can extract examples and they work first time you can't afford to get even a single spaced wrong, so tubes in some other areas solve problem. Although that exclusion, Mugabe.

K

I

K

Our own, an area.

I

Ever known, specific a modules or suggests, we've got three instances of line to know what outlook for was a standard, consistent approach that I in as many places as possible, rather than having listened different. Please.

F

E

F

I agree with you: it's a larger problem, but you know in this we've been talking about tuning a lot right. We.

Y

Want you to mate the.

F

Validation of those examples, and not do them manually- and this is where I'm coming from, where we should just focus on those.

F

Try to make it like.

F

A

Can't as a contributor just responding to your comment, I think we're trying to say is: let's keep it simple, so we can get a solution now faster, but I would say this is a lot more complex than what I proposed.

F

To have this for a yang examples right enough for everything which is like, then.

C

While before you go away before you walk away, come back, do you think this is something that could just be solved through tooling? Do we need a? We need, a document on it? Why not just solve it in tooling,.

C

I'm setting you up here.

C

Of course, we can't hear the black. You know the off my.

X

F

I've been dealing with some of those tools. All these exceptions are cases that you know. This is not the biggest one in the world, but you know just having a convention. There would be way more helpful.

D

V

D

V

D

Useful problem to solve I, don't think she's been a huge amount of time debating lots of different solutions that it is very complicated. Everything should just try pick.

K

D

Quite simple: to do and just move with it fast and get it done around the way.

L

Yeah, more curriculum, I think we should do this.

L

As you read all.

K

These documents of all those examples, if.

L

Everyone comes up with their own.

K

L

Of trying to solve a problem, if it just becomes more difficult to review and check the examples, the tools and all that so having one thing we're doing, it makes sense. I also.

E

Good with camp but I.

L

Think that this can be fairly simple, syntactical change. You choose the context dependent at.

K

L

Should just be a very simple.

C

All right, so how many people think that this problem is something we should solve in the working group looks like a reasonable number. How many think that this draft is a good foundation? A good starting point for the working group, I would say about the same addy curiosity, how many people have read the draft slightly more than either of those. So that's actually a good thing. Often it's less.

A

C

How many people would like to talk and tell us why they think it's not a good idea to start with this document.

A

Does a contributor and I already did I just think it's too complicated.

C

A

In mind it's a starting.

I

C

Not saying that this is the final solution, it's saying that this is the framework we're gonna be working under, so is it and is it a reasonable enough framework to start working with.

C

Okay, so Charles.

R

I sort of owe my comment was the same tense. I believe it just seems, I'm afraid that anything that we write enough like starting.

K

With this, it's gonna be too.

R

Complicated I, just like the very simple you mentioned, the contentment.

K

R

Mentioned that's what I've always done. That's kind of like Unix commands what you do, and so some have so it just seems like there's a clear solution. I guess, I haven't studied it enough to know what other solutions people are doing, but certainly introducing additional white space to make it look pretty on I, wouldn't.

A

R

That's a lot of Greek, okay,.

C

So I think it's clear that we'd like to have a solution in this area, perhaps come back with what you think is the simplest form and check with Charles and Kent privately to see what they think before bringing back to the working group. Does that work.

V

Not you yeah that.

N

Worked for you can okay thank.

A

You, okay great, so the next person presenters blush- and this is the last presentation. That's on the agenda that that's posted, but where you as mentioned the top we're gonna, try and see the time for Alex plan to present his draft and then also maybe time for Igor to percentage.

G

Well, I drank from Erickson again. This is a bit of a draft that is very strong.

G

So there are number of capabilities of the server on unchanged notification capabilities. Just one example of them that are available via yang modules, IETF Ganga library comes to mind, contains all the modules revisions features whatever that call for metering also contains similar things. The C camp group also proposes a alarm module that will contain server capabilities.

G

Most of these capabilities are actually fixed and implementation time. So you don't need the really denote to know them. Some might change, so it is reasonable to have them as in the yang module, but we need to know this information already in implementation time. Nms management system development wants to start before you get actually the node in your hands.

G

Operators like it. That's a waterphone base their purchasing decision on things like will you support radius feature in your note, so they want to know what features and what modules will be there.

G

So, let's document this and as automation and tools will need to read this and anyway, that's a better way to do. Let's do it standard yang, based format and that's where I proposed to have yank instance, data files formalized. So this is data that could be fetched from a yang module by restaurant and that form or another way, but we actually document this in the file- that's very similar to what you would get for a get operation, in this case yang library, where we document that radius authentication is supported actually, while documenting server capabilities.

G

There is the reason why I started this there's another strong use case for this pre loading data when there's often defined before data like before before user groups and the access through, so that site I define that in before. If you have a read-only operator security operator and the system administrator, what areas would they be real?

G

What areas should they be able to reach according to access control, and then you could document these in a file and sometimes even though this file into your system now this data may be stable, may be static in a way or maybe they can be changed later on. The boat are possible so both for documenting the capabilities and vote for pre loading data. We need instance data format. We actually have that, so we don't want anything new base.

G

It don't base it on the get retire formats for both net confer and there's both add some metadata, because, yes, if this is a set of instance, data that belongs together in a wait and should have a name revision, few other things and also sometimes like yang library or the other, our model, it will actually be configuration for status of state data.

G

So I try to define or reuse more like two formats, one for XML put an instance. Data wrapper around it just to make it a single root and also to carry the metadata like the name of the instance, data set revision, description, potentially other data bits, and then you get what you would normally see in a Yank library which very nicely defines actually what your yang server will be able to do, or the modules deviation them other stuff.

G

Similarly, you could have a JSON format. Happily again, this is predefined. For me, the only thing I did was put an instance data record around it and if I stated that this is a gastric tie, I would get four rest comp data, at least that's about how I understand it. There was a proposal that you should be able to define the root resource somewhere that maybe in metadata and then start your instance.

G

Data from that point can be done this we agree and in this case I don't see the need for a tags or timestamps, at least, though, on the individual data bits you have one for the instance food instance. Data set I.

G

Defined in my draft one yang module, it doesn't really contain and it's any configurable data, just these metadata annotations for the instance data following the yang metadata draft. These are the metadata I thought would be useful. We can have more or less and what I see if this is needed, because I need know at least two implementations that are already using such instance.

G

Data files, not exactly these formats, but slightly different formats and but based on these principles, and it is needed for yang push to document the capabilities and we actually needed for yang library and the Al'ar model as well, because we need to know what what will be there development time for MMS and others now a few things that I don't want to do so.

G

This is a big idea to document it document, capabilities with instance data and gives you the instance data format, but it doesn't stay that which exact yang modules you will have to use this for I. Have this other draft that will prove that proposals to use it for not notification capabilities will have to have proposals for any other thing and then the other yang module.

G

It doesn't space that okay, you have this file. Does the toolkit have to load the file as is or will it just generate it? Somehow, that's out of the scope I state that this will be used in some way and also the lifecycle weather.

G

Let's say: I I have a instance data set about yang features: can they change? Can they not that the separate documentation set? That's out of scope again for me.

G

And in the long run, I would like to have this group.

D

Go since this case, so I just had a couple of course I like this work. These useful comments in terms of your model definition you used narrator rather than I, wasn't sure why using it takes about just regular leaves here not clear to me that, because.

G

This is actually metadata about the instance data set. It's not real data that you will load anywhere at this. If you have a look at it, the metadata instance date, the name of the instance, data derivation, of instance, data. It's really metadata about it, and also it gives you a nice encoding both for XML and JSON yeah.

D

G

D

Just put it in the metadata container and there's regular things, then embed the rest of the information lower down might be easier than me. I'm.

K

Not sure metadata helps you go is.

D

Required it makes it so more complicated, I think the other comment, I was just to say, was you've defined XML.

K

D

Format, I think also again doing a binary encoding like seahorse again, could be useful, I'm, not sure so much as a standard license than a relation theme where or an interracial thing, but way to store this data device files.

D

So this may not have to do that. I think has argued.

G

There is no current the binary format. That's one point, I see no problem with defining like their binary format. On the other hand, I think this is. It is good if it's readable, so these would be most more important for me.

O

As part of the gang catalog work, we built a model to first. Yes, we there's value in this. We built a model specifically to do this type of thing. We did it slightly differently in that we refer to the yang library output, but we have kind of this metadata around it. So we.

K

We think of things like.

O

The platform, the device platform, the device version so that you can say this platform with this version- supports these modules, but it was just the additional metadata in your example, but but it we find that very useful for pre loading into the catalog. So we can do some of the things. Benoit said where it says, given two libraries of modules here are the differences and here's what this platform supports in this vendor. That's our use case, so there's I think some good cross over there. I would.

G

L

G

Tags is also one place, one more use case where we could use this.

L

Movement, yeah and similar comments as Rob regarding your use, the job and the patients, and also the top-level container here- is this data. It's not defined in the right. So now.

G

It's just defined.

L

G

In my junkie, we're using something.

L

Like here yeah beta extension, you define it in the I'm module and then you get the you know c4 and Jason, or they read instead of having the top-level thing special and today.

G

I'll look into it I.

U

Hope, that's a as a follow-up to Martin. An adoption is to define this instance data as a normal container, and the new scheme amount for inserting or building.

K

This will be some.

U

Just some snippets that that are not supposed to be validated by and.

G

U

Contents of this, because it.

G

Would be nice to validate them against the individual yang, more use by you think scheme, among other.

U

Ways you are including stuff that that it's not expected to be inserted in such a container, so so, but I think the key muhammd impact can be yeah that you know right. That definition would be maybe.

G

U

The data model for this other than that I I also think it's it's useful work. I guess it could be used also in RFC's for describing some more complicated data models. It could be used by tools. My my to actually use this library, I'm I'm using adjacent formats of this encapsulating element is not needed and it's still useful.

G

For carrying the metadata please, but I can, could we discuss the scheme amount later of.

L

Might also make it a bit less flexible, then you, you know you have.

K

L

Hope level stuff in here I think that this can be with like snippets of data and.

G

Then you would need.

L

G

Somehow, what's mounted inside.

L

U

Depends actually uh I didn't mind.

K

I didn't mean scheme amount in the.

U

Cabin flora, I I, meant the scheme amount. The future part.

K

U

Scheme a month we are going to talk about.

G

Offline, so can't.

A

As a contributor, I agree with Martin I think it's that well one useful work, but I also thought that yang Dail might have been a better approach for approaching this, and maybe look into that and then separately with my chair head on I apologize. We have run out of time. We only have one minute left in the sessions, so we won't be able to hear the presentations from Alex, Flinn and Igor, but let's continue those discussions on the list.

C

We have another session tomorrow. It's right now fully booked if there's time at the end. We certainly can try this. If we have session time we'll use it yeah right, that'd be fine.

C

So as to your presentation,.

K

C

Think you had some comments that would should lead to at least another rev, so I think it would be good to integrate those comments and come back to the working group. I think there's interest in this problem. So it's not go away. It's adjusts the document a little bit towards the directions which.

G

You can refer the patient, yeah.

C

And maybe maybe after you adapt a little bit, though it'll be worth pulling okay with that we're actually exactly out of time. Thank you all very much and we'll see you tomorrow.

G

You seem pretty cool asking this tomorrow: I, don't I, don't think we need much time at all. I've seen this. We don't even need to think it's really just.

B

G

Be there I'll, be there.

Y

You have five minutes.

G

I'm hoping this.