Continuous Delivery Foundation Events Special Interest Group, 21 Feb 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CDEvents Working Group (EMEA/AMERICAS) - Feb 21, 2023

Description

For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/

A

B

A

Doing hey I'm doing good nice nice.

C

Yeah so making a few edits uh to to the CD or the uh connecting events uh proposal.

C

Yeah um I added the one use case, uh I thought that was really important and then I also added a hybrid approach section um because we talked a little bit about that and I think I. Think a hybrid approach might be the best solution, but we can talk about that once uh once we get started.

D

A

um Some more people join us.

B

A

Andrea will not join us today in some detail.

A

Expect another minute I see more people trying to.

A

Okay, I think it's about time to start I can share my stream lists.

D

A

So welcome the steel events meeting today, wherever transverse, please sign up on the notes, uh so I have more the same agendas. We usually have I've added the the ongoing items as links there and then you're free to add whatever you want as well. In the end, uh let's see about how we can proceed on this, so first just look into the the roadmap.

A

Here we have it as as a reference link mostly uh we last week, we should have had the release we had planned originally like in mid-February, but we prolonged it now to the end of February I'm, a bit concerned, as you might understand that we might not reach this either, especially now since Andrea is on PTO this week, and we have a lot of things to to cover still which are in to do.

A

Most of them are actually often assigned to Andrea some or not, but there is some work in order now, I'm pretty sure we will not reach this um yeah, we'll probably come back to the the sdks a bit later on as well. um But as you see, there are 100 tickets ongoing and some are not really started and I've also created some new ones. Some some bugs have been created here.

A

Some of them are my bolts for creating, uh but anyway they need to be done so I, don't there's nobody really need to rush this release. We don't have any such fixed deadline, but require us to to release it this week or next week.

A

um The thing is that we want to have released close to the Silicon, so we can and show off that uh the most recent things when we are there for something so it makes I, don't know it might be so that people don't do something then, but I would would hope that we can type a CO2 release now and then a CO2 release goes to Silicon and we'll see how it works out.

A

Okay, so we had some action items uh from previous meetings. uh One on Andrea had to create an issue about the former release process. I haven't seen, but we have such an issue.

A

I, don't think that is done, and then you said yes before that you are about to write some use cases, and you have already done that yep um so that we can I, don't know. Do you still do you want to keep this as an action for coming weeks? There should be no.

C

No I I wrapped up. There was only one really big use case that uh was I, think important to capture and and I and I wrote that one so um so yeah that should be done. Yeah.

A

We can consider that we can come back to this way. We are there and then Andrea had a close connection on the connecting events. Part to add some details about tracing context and I. Don't know if that's done that, so we can.

C

Let this data when.

A

We come into it yeah.

C

I saw some changes uh to the trace context, so I did I do think he did make some modifications to it. But I don't know you.

A

Know something I.

C

D

Know if you made that much.

A

That yeah I'm not exactly sure what was the definition of dance. Let's say for that action, so, okay.

C

A

Enough, maybe this is good enough. Maybe he can answer them back when he's back next week or next time sounds good. So if we just take the different ongoing items in order, we can look into the incident Events. First, there has been some discussions in this we'll request back and forth. I made some comments late last week and I'm pretty sure, Andre I didn't have time to look into those before he left for holidays.

A

So there are some things to be done here still has also requested changes, I'm, not sure.

A

If there's an outstanding actually needs from him, then, but and that's about it and I had some questions something so it requires some more work on this I would say, or it can be there- and this is maybe the main reason for the CO2 release as far as I can see, at least that we want to have it as soon as possible, uh so because this is what lacks to be able to fulfill the Dora metrics for core volumetrics, because we really need this for that release. So it's there's no really need to release.

A

Cd runs without typing this in place, I would say, um then we can go to Connected events, then. So, let's see we are here. This is again the document where we discussed this. We don't have any. uh Maybe we have an issue for it, but the way we have at least until now discussed this through the again the document.

A

I think that's, that's perfectly fine.

A

um So let me just here.

A

E

Hello, my friends, how are you uh yes.

A

So Ben, would you like to list sure.

B

C

um Yeah yeah yeah. So if you go to use case so I added this uh last use case, which I think it's kind of like the end like this is what I see people doing with CD events and the kind of like in the future like not like the immediate future. But um since CD events, you know is about interoperability. You're going to have CI CD systems are gonna, be you know, kind of PL be able to be plugged into various different other CD CI CD systems right these individual Parts could be just plugged in.

C

um However, every one of these uh systems is gonna have their own UI.

C

So it would be great if there could be a singular UI that understands CD events as well as being able to make sense of everything by uh you know, using links, for instance, linking these events to show connections between all these systems. So I think that's where the biggest benefit um would be had here is is this? Is this use case all right sit down, yep and then uh added one section towards the bottom?

C

It's still pretty uh I'm not really filled out, but uh the other section I added was a hybrid approach which I think I spent a lot of time. Thinking about this and I think I think this is going to be the best, the best approach. So the reason why I say this. So if we go with Trace context right that adds a huge dependency uh for for CD events and I. Don't really like that, um and what I like about Eiffel. Is that it's very um it's it's.

C

You know it's small and it's too it solves one particular problem, um I mean so does tracing, but tracing does more than what we we need it to right, like you can add, tags and whatnot, which you know I feel like CD events would already have with like custom data and everything, so it's kind of like being redundant there, so I think keeping something simple like with Eiffel and then just utilizing that Global ID that we talked about uh last week.

C

Email would be the perfect solution and this kind of solves both problems um with you know the the fast lookups, but also giving you the graph capabilities, um so I want to flush. This section out more, like I, want to show like what I think this should look like um and then I think once Andrea is back, uh we can do. I can attend the next, the next Sig meeting at 5am, and we can just finalize a decision, um but I really think this hybrid approach is going to be.

C

Is the best is going to be the best one going forward um so so yeah? So that's, that's kind of like where I'm, where I'm at because, like I said like the last like week when, when we first when we talked about that I've been thinking about, you know what the benefits is and then I started thinking. Oh, we could use Eiffel, but you know like if we use Eiffel by itself, this UI use case is going to be really expensive.

C

Whenever you know you have like thousands of users trying to you know, look up, you know like what their flow look like. That's gonna be really expensive, so um I was like okay. Maybe we can still do the the global ID thing with Eiffel and then yeah, so that's kind of like where I landed.

C

um So it's been kind of an ongoing process with with the whole thinking but yeah yeah.

A

I think it's an interesting thought: I'm still a bit uh uncertain what you really mean with the global ID, where it will really be instantiated where it's been degraded who like who has control of when the new Global ID is created, and such things.

C

Right, right and um and I think that's definitely a fair question, but I think we should lay the leave this up to the users I think of when those things are created or not, um for instance like when you use tracing. um So this is kind of like piggybacking on how pricing uh solves this particular problem. Is they leave it up to the users? So um they do a best effort on trying to guess um what you want, but I. Don't think we need to go that crazy, I think we can just say you know.

C

Maybe it's just um if you want a new Global ID, you just call this one function and you're done or like um maybe put like in in Java. Maybe it's just like an annotation that you do like anytime. It enters this method or whatever it just instantly creates This Global, ID, I, think having it be driven by users is gonna, be the best bet, because you this gives the ability um for a UI to be molded and shaped to like what someone wants to view right.

C

um So this part, letting the users drive, that versus us I think, is really really important and I think trying to solve both of those is going to really complicate the solution, so I think just having the users Define. That keeps things simple um and- and you know like like how Trace idea, how tracing Works um and and gives uh gives the necessary benefits of of the global ID. So that's.

A

Kind of like what I'm thinking, yeah yeah, of course, I mean how to create a globality. Might not be that the thing that we should decide on here. That's not what I'm but I mean. So, of course, there could be different ways to create an ID. But I was more thinking.

A

If you have a like one step in your pipeline, that is totally triggered by an event yeah and not like not uh orchestrated by some some high-level python orchestration engine say, for example, you have a testing process that you would like to kick off whenever some artifact has been built somewhere right.

B

A

So then, that testing process it could either take the global ID from the triggering event that is in the artifact created event, or it could take it from some other location where it knows that well, I'm part of this flow, so that price should my coming. Events should always have this Global ID foreign.

A

The the Upstream events outside into the new ones.

C

Yes, my thinking was always Upstream. um Whatever is triggering yeah because, like that I think that makes the most sense.

A

um And then, if you would have like one test activity wanting to run either this or that artifact has been created, for example, it's two different artifacts. Maybe that is being deployed in the same environment and then those artifacts might have different globalized than in them.

C

Yeah but those would be two separate triggers, though right.

A

No, it doesn't need to I mean it could be just an oral relationship if this event comes or if an event comes with its artifact, ID and another event comes with this artifact ID, then we should start this testing.

A

Not uniquely identifiable what exact, Global ID to always use.

C

Right right, yeah.

A

C

No I see your point. um You know because, like you're saying like if these artifacts were grouped somehow you'd like to like those things uh grouped um it's kind of like what you're getting at, um which which is completely Fair, uh you know and I think honestly.

C

um The question is whether or not that's important um it pro. Is it like I, don't know. Let me let me think about.

A

That I I might be the colored by the way of thinking and I feel better, because I think one way to solve that I guess when I think with myself. No, it's like you could have like an s-bomb or something saying that this is now the the Baseline of this uh test environment that I want to run some tests in and that test environment is. Has an I mean an s-bomb of these two input artifacts or something like that. So then it's like the s-bomb object creation.

A

That's that's the maybe the global ID for the continued process in this pipeline, or something like that right right.

C

B

C

Thinking about this, like from the UI, the UI use case and I, think it makes sense for the for the Upstream thing to trigger it, and that would be its own Global ID and then the UI. If it wants to go one extra step forward, could maybe utilize um like links like actually the graph traversal like if it wants to make that expensive call to relate all the artifacts. Does that make sense.

A

I guess so, because then, if if there is I mean a new Global ideas that say created somewhere in the middle of the pipeline, but there's still like links connecting these pipelines to each other, then the visualization or the U right there could could follow those links in that case and the global IDs in other cases. Maybe yes.

C

Exactly exactly so, yeah I think, I, think and I think that's why the hybrid approach works is because, like it definitely like links is like a good for, like you know like if you actually want to see like relationships between, like you know, like all the events, especially with these crazy, like forks and whatnot, but with the global ID.

C

What's nice about that, is that quick, lookup, so I think like when you want more information on how things relate, then you utilize with if you go with Eiffel right, like especially like when you know things are going to be detached like, for instance like when artifacts you know, because you lose that grouping. So when you know that that grouping is going to be lost, you know, then, then that's when you use Eiffel versus a global, ID um and I think we can state that in the in the hybrid approach right.

C

um So we can come up with these use cases where uh it's questionable and then answer that and then show that in diagrams, as well as like what the the payload or you know, what the model or spec would look like um and I think you know that can be an action item for me.

C

So I can flush this out more um with with that particular issue that you laid out, which I I definitely see the issue there, but yeah I think I think this definitely could be solved with uh with just using utilizing either Eiffel or or you know, uh or Global ID choosing, whichever one you want to use to uh to look up um the relationship between things.

A

Yeah I'm thinking about the use case, which I believe is installed by by Trace context or a parent-child trace IDs. Somehow.

B

A

Not sure if you've seen this before from the honeycomb uh GitHub GitHub sites- and this is where, as I understand it, it's only using Trace context with parent-tech relationships, yeah and it can anyway distinguish from parallel and serial activities based on I. Guess then the end timestamps of these activities. So if the next activity is started after the previous one has finished, they are assumed to be serial.

A

If the next one is started within before the previous one was finished, then it's assumed to be in parallel executing with that one I think that's how it's interpreted and then I knew I could probably handle it. That way, and with that, we don't really need the links to show this or to to visualize this scenario of a like a top level Pipeline, and then it's different steps underneath it both in parallel and serial.

A

So that's also a way to handle the the uh should I say, maybe not the relative by an internal scenario, but at least the the progression through a pipeline. Yes,.

B

Right right, it's.

A

Not really possible to say I mean, for example, if these two things here different builds, for example, and then these two are different test activities for the the corresponding bills. It's not really easy to see that the first one corresponds to the first build and the second test corresponds to the second build or vice versa. You can't really say that from this just having a parent child parent child relationship, I, guess, then you would need some kind of Link saying that.

B

Yeah this is yeah, it.

A

Relates to this actual build activity and such things yeah.

C

Yeah definitely so yeah I'm, I'm, very familiar with uh yeah tracing, um so yeah I think I think what you're saying uh makes sense and that's exactly how tracing works is that you have this.

A

One sorry but I'm trying to get at it's it's. Maybe we would need uh also this kind of parents. Child relationship in these ideas would that be any benefits you think to have possibility to like spawn off a Child ID.

C

um Yeah potentially, um but I want to keep this, so my whole goal is like I'm a very much an iterative um type of type of person, so like I, want to get the uh kind of like the foundation good, and you know maybe the parent child could be in that in that Foundation, but I think that can be an additive feature uh later.

C

um I just want to get the uh you know like just making sure that it's you know. Scalable is my number one thing for me um as well as being able to be able to visualize these things in, in that graph, like as an end, consumer um and and definitely I, think a parent and child ID definitely makes sense.

C

um You know if you want to do even further grouping, um but, like I said that can get pretty crazy pretty quickly. Actually, when I was working with tracing and I got that got very complicated very quickly uh and like there's a lot of places where that could go wrong and actually a lot of our bugs have come from uh creating bad children uh and not relating them properly. So that's that's a little bit of my concern with introducing that, uh especially now.

C

uh uh You know where we don't know if we need it just yet. However, if someone starts saying, oh, we actually need this feature. um Then I think we should probably add it. However, for our particular use case, especially for this UI one I I, don't need it um so I think um I'm happy to leave it out and then, if we need to add an addendum to the proposal later, we we can always do that. How do you feel about that? Emil yeah.

A

That's tough to find we can start here and then, of course, we can elaborate on on child IDs and writing custom data, for example, to begin with, or something like that. Yes, if we need it or not, yeah.

D

A

Out another idea here or another thought now: you're talking a lot about the UI part there and making sense of the data that comes out of the pipeline as a UI I mean looking I would call it like a polio commit when you see what happens with your commits through the department, regardless of what tool actually emits those events.

B

Yeah but there.

A

Is another use case which is not urinated, but this instead AI or machine or machine learning related that you would want to make sense of a pipeline and maybe predict what should happen next step in a pipeline and such things and and what tests should you perform, because this kind of commit that comes in open uh Fades?

A

This that's the case down there or something uh so does anyone of you in the meeting have heard any experience or any connections to like AI people who would have some thoughts on whether it's it's beneficial to use this explicit links between the events or to use Global franchise? If so,.

C

I could get some information from from AI people if, if needed, um actually I have a meeting, uh not this week next week, with someone uh from machine learning uh at Apple, that I can I can ask them, um but what's what's a particular just like which which they would probably use or what's what's the question yeah.

A

What what model of the connecting events would be most beneficial industry? Okay,.

C

Cool yeah, yeah I can definitely ask um yeah. Let me let me write down all the stuff. I need to do and then.

A

I I don't know I think we had like an ml stake before is that is that there still or is it? Has it been swallowed up by some other work or just hundred percent? Do you remember, didn't we have an ml stake somehow Insidious.

B

Oh uh yeah I'm not too familiar with that thing.

C

Okay, but yeah yeah I'll I'll get some information. uh It's. The meeting is like middle of next week, so I'll probably get some information uh towards the end of the week uh and then I can update or I can at least Post in the um in the slack Channel like like what they would prefer and whatnot or I can add it to GitHub issue. Whichever is preferable um where these discussions, um um or these ideas or contemplations um would be best. um Yeah I can put it wherever appropriate.

A

Yeah yeah see we have an envelopes Sig. Actually it's still uh well. This is a selection record. It seems to be something oh.

B

D

A

um Might be because I think it they people who are more involved in normal would be more. We could have some valuable input to this discussion. I would would guess.

B

Yeah definitely.

A

um Good, so I didn't really write down any notes for this now, maybe I should have done that. Okay.

B

A

But did you say you would.

C

A

C

Fill in what I was supposed.

B

To do here, let's see.

B

And then okay, let's put.

A

Coming down from there.

A

Maybe I should have an action to reach out to ml folks about this. Yeah I can take that action.

A

Oh you yeah. You have your email people in an.

C

Apple yeah yeah we can. We can both ask you.

B

Yeah I think that works.

D

A

Good, so anything else, we should say about the connecting it on this discussion then, and.

C

Now I think I think we're pretty good um I. Think we're kind of like getting towards the a place to um you know: choose.

B

Especially yeah, especially.

C

Like with with this hybrid section, I think this is like I said: I have a feeling. This is what we're gonna, probably end up, choosing, but definitely getting like I said getting more information from other people like about their use. Cases I think is, is going to be.

C

um You know really helpful in determining that, so so yeah.

D

That's good, that's good! um So.

A

I think we can leave it at that time. Now uh then, for the testing events uh there is this pull request created by some time ago. I made some comments on this as well and I. Don't think those are handled so and I think always also away this week.

A

So we're expecting to come back these comments next week or whatever um we had last week, a presentation from Bruno on what test cube is about, and if you weren't there you can listen into the recording afterwards should be on YouTube. Never did it um so I. Don't have much more to say about these testing events again. I think there are some some things remaining here to be sold before this can be merged.

A

um That's for that, then, you can just visit the tools, integration board that we have.

A

I, don't know if there's much that has happened here really and the Java SDK has been merged. Now you can see so there is a job SDK to be used. I think this is yeah. This is done for 0.1 release, of course, of the CD runs.

A

So this cannot started to be used by by Jenkins box, for example, and by Spinnaker.

A

um That's good Jake and six. We don't have anything there mark just drinking sex in Spinnaker. Maybe you have more information there than on the progress on the integration yeah.

C

I just added a section on that, but yeah.

D

C

Yeah, so I was just gonna.

D

Okay, yeah yeah.

C

Okay, cool cool, um so yeah Spinnaker RFC, uh so julender just made an update a few days ago um uh to to the RC so I. uh He asked me to look at it, so I'm gonna take a look at it, um probably sometime this week, um so I think we're at a pretty good place with that.

C

um So once I give the review, my plan is to actually talk to some of the um Spinnaker folks to see like where you know like how much more do we need to flush this out before we decide to accept it um where I think. As long as you know, he addressed uh most of our comments. I, don't think it's gonna be too much of an issue to to get that merged in um but yeah, because I I think uh the biggest thing was just you know referencing.

C

um So one of the biggest things is uh talking about the implementation details in in Spinnaker so before um he was just talking about the overall design, but there's a particular microservice that we wanted him to mention, and you know utilize and document, um and he uh he it looks like he's, he's made those updates, so um so yeah so test. For me. Action9, for me, is to review that um and then hopefully uh get that merged in soonish.

A

Cool cool looking forward to that perfect.

A

um Yeah, that's the spinnakerberg and then the actually implementation is system that is supposed to get on there. I guess uh I guess this refers to the actual implementation. Then.

C

Oh, it's the RFC, or is it oh, it's actually.

A

Yeah yeah, okay, okay,.

A

And then in tecton I, don't think we have anything, no shipwrites, none Persia, oh arrested scale, yeah! That's the perceptual I'm notified when it's volunteering to start working on the real estate, SDK.

B

A

There any CI TV system that uses.

B

A

B

Oh I didn't even know that interesting.

D

A

uh There is second recommend the uh what is it the text from TV show about Persia that was launched in December I believe last year, uh where it was discussed, and also the one of the head maintainers of of rust itself was there in that in that screencast yeah, okay, it was quite a good option.

A

Then test cube is what we talked about where we had a presentation. Last week.

A

And the testing events starting over there- and this is afterwards so yeah- it's the status of the ongoing activities and there's been a car receivable Dimension when it comes to the SD case. I can bring up the board here again. This one and I had contact with the maintainer or the author of the Python SDK Eric and unfortunately he has been very busy with other things. He hasn't been able to prioritize this and it seems he will still be for a couple of weeks.

A

He is about to change jobs, it seems, but I I asked them if he will be able to work help out with this after he has changed jobs, and he said that they are. His new employer actually was very fond of in contributing to this. So it looks promising. So we we should hope that he can come back in more strength to to continue working on the python SDK.

A

That's unless someone else in the City events Community takes on this I guess nothing will happen on it, not much at least in the coming few weeks or two months, even.

C

A

C

um Since we mentioned uh arrest, SDK I think now, though, since whenever we start the rust road, if you will for that SDK or for whatever new SDK I, think it's important at that point to use a schema to generate sdks, um because I've noticed like anytime, we made change to a spec. It's like you know. Andrea has to make all these changes to like the go SDK and it's so nice to go, update the Java, SDK and so on and so forth for the python, SDK I think it'd be really nice.

C

If we could just change it in one place, have that be generated, and then you know fix any of the um compile errors that you know that introduce the new features or what have you right um so.

A

They made like a code generator with this team up as a base, yep yep, yeah, yeah I'll show you that's really existing yeah a little bit well,.

E

We we yeah, we struggled with this at liquibase and uh what we have is um you know basically a prototype um that it's a structure that we use to generate this stuff. For you know everything from java, SDK GitHub actions, those sorts of things, so the person that's in charge of GitHub actions would go to uh that. You know that prototype and say: okay, what's changed and generate their code, so each SDK is utilizing uh a kind of single source of truth. That's how we've done it at liquobase.

A

um What programming language should I use in there, then for the sdks.

E

um Well, it's all well, we've got. um We certainly have Java we're using it also internally for all the tooling that we built that calls core liquibase. So that's a lot of java, um but it's for other things like I, said GitHub actions generating those yeah.

E

um Let me go ahead and see. Let me go ahead and pull it up and see um and get some more details on it. I'll share with you.

A

I guess it would put some requirements on our source. Then I mean on the on the Json files scheme on themselves. Yeah.

E

And exactly- and you would certainly you can you'd- be very naughty to change it there's a commitment um but I where we've had success. Everybody at liquid base said wow, that's a great idea, but it was always like number four five or six on the price priority scale and what really pushed it over was that somebody took it and implemented it for their stuff. They were in charge of a specific API I forget which one and they did it and they got it done in record time and uh there's there's a little bit of fomo.

E

The other teams were like well I want to use that so by having an example of that, uh let me check it out real quick.

C

Yeah I think to scale um like as CD events like, inter, like as a as uh an org like you need to generate like you're like you, like, you said you know like python. Your guy has been really busy right. That maintains the python SDK, but if you guys could just generate and fix the you know like compilers or runtime errors, you know, that's that's really simple right. It.

A

Sounds easy enough, but I'm not sure if it is easy. Yes,.

C

But no no General the generation portion is going to be very yeah. That's gonna require some thought. That's definitely going to require some thought. It's going to require um some. You know some technical prowess to to do this, but I think spending the time now banging your head against the wall will save you much more much more time later.

A

um It would be good to see if there are any such uh or what, of course, there are them.

C

Yeah, open API is a pretty big one um that I'd recommend um there's also Smithy, that's also pretty it's it's used heavily at AWS um and then open API is also part of swagger or Swagger is a part of open the API. So there's also also that um so yeah there's definitely a few tooling. um There.

C

I think it's yaml, but I think you can provide Json. If you choose.

A

C

Yeah and then open API is really pluggable too. So, like you can um add your own custom things to it.

B

um Heavily uh it's one of add some notes. Have we used in AWS, okay, I used everywhere? No.

A

I know what that is: I'm sure, but I haven't really seen that it generates sdks, but maybe that's it.

C

Does yes, oh yeah yeah yeah, so it does um typically so uh so at AWS, when I was there, uh we used our own internal um thing, uh it's called Coral, um but eventually we moved to Smithy, um but some teams were using uh open API. So um you know uh yeah, you can totally generate sdks and some even open API has a SDK generator on top of it.

C

But you probably don't want to use that or you probably want to have your own custom plugin in that and then utilize that tool, but it does have its own custom. Sdk generator.

A

Because I remember Andrea said in in the beginning, when it first created the initial Json schemas for proceed events, he actually generated the schemas from the go SDK itself, so the other way around them.

B

C

A

Can also do that. Yeah yeah, and maybe that's that's the first thing, but then, of course, if it's easy to generate sdks some different program, languages based on the schemas, that's even better yeah.

C

Okay cool, so yeah, so open API does support Json. So it's Json or yaml.

A

C

So may may need to do a little bit of tweaking there, but you know it's definitely definitely possible to do to do this, but, like I said it's going to require a little bit more thinking. But um I feel like we're now at a point where we have a few sdks, where this actually could start being pretty beneficial.

A

One thing with with Jason: obviously there is a limitation in the Json format that you can't really have comments. Yeah.

B

A

Right I mean if you would generate a well-documented SDK from a Json schema uh specification. You will not have any documentation at all in the SDK. It would just be wrong.

C

I bet you it's a separate field or something like that. Like.

A

Maybe instead of course, then it's possible to have yeah whatever you want so yeah, but that's an interesting thought. Strange I haven't really thought about that. So much before that, yeah I thought it was not really feasible, but maybe it is done.

A

um I bet there might be some kind of blue layer on top of it, so that we would need to somehow update, but not done for each version.

C

So yeah yeah, so uh so what these sdks would generate is basically a very rudimentary.

C

um View or you know like apis, if you will and then what you would want to do is to like you were saying, is ADD glue or make higher level abstractions to make it easier for users. But, like I said you know, if that if the scheme was changing and whatnot, you know that you don't have to change these abstractions right. You might, or you might have to just do very little so um so yeah I think this could go, go a long way for you, guys yeah.

A

I think it's a second is very good because it will also put good requirements on the the students.

B

D

A

Standard and we cannot really deviate them exactly.

C

And to make sure all the sdks are yeah exactly on the same spec, which is really really nice.

A

For sure for sure yeah, that's a very good thought uh and I invite anyone some time and energy to spend on it. But to look into what options there are uh I've.

C

Actually done I've done a lot of code.

C

Generation Um I will have to ask my manager, though, if I can do any work on it like the problem is like anytime, I touch anything code outside of Apple like, except for like approved things um where we're still trying to figure out like the whole CDU events with with our team um I need to ask, but once I get if I can get a yes or no um I'll, immediately tell or I'll post in the slack whether or not I can work on this, but if I can get some time like I I can absolutely knock this out pretty quickly or quickly.

C

A

So I'm, not at all against I, mean, for example, moving from Json to yaml for our schemas, if that's a better option, of course, just to be able to make better go generation or SDK generation.

C

I'm I'm I think we should move to yaml just to have com I mean just to have comments in your spec. To be honest,.

A

But that's that's one thing and we actually did the the same uh thing in Eiffel recently, we actually moved from Json to Yamada since the core, uh the the schema, this game aspect for us, you know, because there we also don't have the documentation as part of the schema. So we generate Market down documentation from that those young requirements as well.

D

A

So that's uh I would Embrace that as well, but it requires again some some attitude.

B

A

Of course, uh good thanks Robert for asking as well. We look forward to the updates. Oh.

E

Yeah I wish I had a definitive answer, but.

D

I'll, let you.

E

Know when I get one.

A

Yeah sure good great, that was good input, I think and good ideas, good thoughts, so I don't have anything else planned in the agenda. So it's up for any comments or discussions. You would like to bring up. You have any.

E

Ben um there was an update on um the spec for CD events in Spinnaker yep. um You were on the meeting, yes I I. My update is that you went to a meeting, but anyway, just is there anything. I can help with that um like like it seems like it's going really really well. I just wanted to get your take on it. Yeah.

C

Yeah, uh so things are moving along uh well, so jolander, the one that uh is the author of the RC uh just made the necessary changes um that we had asked uh from from the meeting that that you had just mentioned.

C

um So it looks like he addressed those comments uh that was done just a few days ago. So I haven't had any time to review those comments just yet so my plan is to review that take that back to uh and then you know, ping a few people that were also interested, and that were also at that meeting um to see like where we're at on. uh Potentially you know merging it, um hopefully there's not too much more work that is needed, um but yeah, that's kind of like where we're at right now.

C

So this is more of me reviewing um and then taking that back to Spinnaker kind of discussing uh where we kind of see uh the plans are for this RFC.

E

Okay, um I'm definitely not trying to Horn in on it or exert any opinion on it. The the reason why I'm asking is because just remind everybody um from Toc perspective, I very much want to go to other projects and say wow, look at what Spinnaker did um and my goal is to have other projects do this as well.

E

Obviously you know Ben I, know Jenkins and Spinnaker are near and dear to your heart, so certainly concentrating on those um and having some kind of um you know you or someone else talking about the value of it, what you're getting out of it, the benefits um and I think those two things having Spinnaker and Jenkins. You know supporting CD events having a user speak about this in in you know, glowing terms that creates you know, fear of missing out fomo and hopefully the others will follow.

E

If more teams do that, if more rather more projects do that, then it becomes even more compelling for our friends in commercial land, um I'm thinking, harness and and the likes. You know, octopus deploy stuff like that, so uh that that's my hope and that's why I'm asking so. Please don't pick up you.

C

Know trying to rush? No, no, no, no! No! No! No! No! No! No! No! You know like, like I said, like any questions or whatever you do, that you have. You know just let me know and then, if you feel like you need a pressure like. Don't don't worry about it, you know it's completely fine, um but yeah. So I I completely agree. um You know like and that's where I think we're we're trying to get it into well. My goal is to get into Spinnaker, so we can start using it internally at Apple.

C

um So you know I think you know if we start showing that.

D

C

uh Companies, like apple, are using CD events as well. As you know, it's.

B

C

Jacob, that's gonna, be a pretty big message, I think to uh to all the other players, so I I'm, that's my hope, but we'll.

E

See yeah mine too yeah and right now, all across the world. There are product managers at CD companies that just got a chill yeah.

B

E

It was just like oh no something's, coming yeah.

C

D

C

Yeah no problem and like I said you know if I have any updates, you know I'll I'll be sure you know. If you're, not in this meeting, I'll I'll ping you on on slide because I saw you were on there and then, if you had any questions we can, we can. uh You know, adjust that that sound good, absolutely.

E

Yeah I'm subscribed to that uh GitHub issue, so I'm watching all the changes. Okay,.

A

Let's see what's going on.

E

um You know just just letting you know what my hope and dream is, you know, hopes and dreams are and what what you know? Hopefully the benefit. You know the TOC can provide um and- and you know getting this with with other folks- because it's all fine and good- that you're working with Spinnaker, but we need to have cd events in it. Just I I I've got a um it it it just it's silly, not having you know, projects working together, I, just you know, it's just a a personal opinion.

E

uh I I hope it's shared yeah.

C

Yeah, no definitely and and I think you know and I. Think too you know, starting you know with Spinnaker will hopefully be the stepping stone that we need. um You know and then I think there, because now that we have a CD tool like Spinnaker, maybe the next Focus like you were, you were saying: Robert is maybe Jenkins or maybe even GitHub actions. You know, because I do have some contacts that GitHub that I could reach out and be like hey. What do you guys think about this?

C

um You know and even maybe even having uh a plug-in that generates CEO events, because I know you can have like different plugins in in uh in GitHub, so that might be an option as well.

E

All we need is one you have one commercial vendor doing this right. They.

C

All exactly yeah.

E

um That that that's you know it it's ten percent. You know new Cutting, Edge stuff. Ninety percent not falling behind from.

B

E

Perspective all right cool. Thank you. Yep yep.

A

So speaking about the case of actions, there wasn't there some initiative by was it Brad? Who I was about to do some gift of actions, integration with students? Oh.

B

I have no idea, but that.

D

A

Trying to find that issue, someone wrote it, but maybe I can't do it now on the Fly. I was thinking also when you discussed there on on Jenkins or yeah, of course, but the other tools I could follow. I know that Jenkins X has a case of initiative proposed where to integrate with City events, I'm not sure their time down when they will decide on what projects to to take honors, though, but we could of course hope that snake is X would go for some. Some such piece of activity during the summer.

A

Good and great, no but I totally agree. It could be fantastic. It combined, like apple, would Embrace.

D

A

To say we we really need this and we we.

E

A

It or whatever well.

E

And it's not just also motivation from uh you know it doesn't just motivate the other projects to integrate with CD events. It's also other folks uh that uh you know look.

B

D

Admire them, they're, like hey.

E

You know you know what are they doing? Maybe we should do that. um This whole value of the you know state of devops report that door was putting out for for years. You know, hey, look at your peers um and and that's really what we do over in CD, land and devops. We we gauge ourselves against others and see how we're doing.

D

E

Hopefully, in theory oh yeah well gang I'm, gonna bounce. Thank you so much. Okay,.

A

Thank you all right, thank you. So um do we have anything else to handle today or can you quit buying them? No.

C

I think I think that's pretty.

A

Much it I think I've emptied my mind at least yeah.

C

Yeah yeah- um and you know, like I, said, like I'm just amp for SDK generation, because, like I've I've wanted that when you guys first started talking about sdks, but I was like you guys only had one SDK, so I wasn't sure yet, um but now yeah now that it seems like that boat is kind of, like you know, uh is uh full steam ahead. I I definitely think this is um something that's going to be awesome to do like I just enjoy code generation.

C

To be honest, so that's why I'm looking for it but yeah.

A

I think it's very cool um just think that they're part of some problems with it, but of course some obstacles could be overcome, I hope, yeah.

B

Yeah, no, no definitely.

C

B

A

It's not yeah, it's not always to see the bullets, but often it could be I guess. Yeah.

C

Yeah um but yeah yeah, no I, think I think we're pretty much I think we're pretty good here so um yeah. If we want to wrap this up, I'm totally. Okay with that.

A

I just think this might be the thing I was after oh.

C

A

C

So I remember Andrea mentioning this actually.

A

Yeah, okay, Mauricio and also uh Brad McCoy has been on to this okay, okay, so I think Marissa was interested in it uh and then Bill McCoy I. Don't think there is a pull request for it. Is it no?

A

It's added to the zero to Hearthstone, but uh I doubt it will be there, of course, but yeah. It would be interesting to see if we have anything they're, responsive, okay, good cool, uh then that's it for today, then. Thank you for.

B

A

Discussions and uh have a good evening all right, see ya thanks.

B

A

D