wasmCloud Community, 8 Sep 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: wasmCloud: New CI Pipeline, 0.50 Update - 09/08/2021 Community Call

Description

wasmCloud is a platform for writing portable business logic that can run anywhere from the edge to the cloud, that boasts a secure-by-default, boilerplate-free developer experience with rapid feedback loop.

A

Welcome to awesome cloud one day, wednesday for september, the 8th, uh we're going to start with a a quick review on the new ci pipeline and then review on watsonflood.50 status, 5.0, and then we're going to go into some discussion around an operator brooks turn it over to you.

B

Sounds great, let me share real quick.

B

All right, hopefully, should be looking at a bunch of just ci looking stuff. A github action is that right, yep, okay, cool! um So over the past, maybe a week week and a half I've been kind of overhauling our github actions.

B

All of our ci cd is hosting github actions um for for what it's worth, and I just wanted to show off some of the new architecture like some of the new things that I added um overall, it made rci a lot simpler and and there's very little that you have to change going from repository to repository to like publish a crate or even to publish like a provider archive or an actor, so I'll just show off a couple of things.

B

They a lot of the actions are largely you know, kind of cut and paste moving between repositories and um I'll also talk about how we're like reducing that cut and paste too. So to start this is the control interface, github action, it's in the control interface repository and it's essentially what uh publishes this. uh This is a rust crate.

B

It's what publishes this to and there's there's two parts of this there's the jobs that we run when you pr or on any new, commit to main and then there's the um there's, the actions that you run when you push a tag- and those are two uh you know- create a release and then release to create.io.

B

So the only thing that runs when uh you create a pr or sorry that if you, if you push a commit to main, uh which we have branch protection on, so everything happens through prs anyways. So if you pr into main, then it runs. This rust check action runs on ubuntu. All it does, is it checks out the repository and then runs the wasmcloud common action for rust check. So this is the first thing that we kind of overhauled for our ci.

B

Is we had this like cut and paste brush check, which was just like a couple steps to like run cargo, build cargo test, clippy formatting things like that, and I've abstracted all those into this common actions repository.

B

This is public. You all are welcome to like reuse any of them. um I haven't actually versioned it. I'll.

B

Probably publish a version one once I've done, like um you know, running through all of the all the different repositories make sure I don't have to change anything, but um all of these actions are essentially reused throughout our github org and I wrote some documentation here, but if you look at like the rust check action that takes just a single input where the crate is located in the repository and then it will check your formatting, for you run cargo build, uh run, clippy and run cargo tests kind of just what I talked about before, um but instead of copy and pasting, this entire thing everywhere and I get hoborg.

B

We just say that we use this action um and so that way in the future. If we wanted to change anything about this specific uh action, uh we can change it here and it'll change it throughout our whole org and as soon as we're not like right now, I'm doing it just off the main branch.

B

We can also do that off of like a version tag, so we don't actually accidentally, um you know, create incompatibilities.

B

So the rest of this only runs when you push a tag, and this is how we actually create a release.

B

What it does is as soon as the um as soon as the cargo build cargo test passes, it'll create a release in github, so that you can, you can actually go and navigate to that release with any notes that we publish and then it'll run the common action for releasing to creates.I o.

B

And if we look at that, one real, quick.

B

That one takes, like your crates, published token, that you can use to authenticate and then uh logs in for you and then publishes your crate so across this repository that we have here now, there's common things that we're just copy pasting across like multiple github actions, and this really helps with with not repeating ourselves in a lot of different places somewhere else that this really helps. Well, I mean still in other github actions, but for like publishing a capability provider. Here, let me go into my pr.

B

Actually that's where the most recent stuff is so for publishing a provider archive. We can reuse a lot of these common actions like we run gnats and then run our integration tests, for example, because we need nats for the provider archives and then we can greatly simplify the process to like build a provider archive by saying just a just a one-liner to install wash or a one-liner to install the cross tool that we use to like cross-compile for different architectures.

B

Things like that makes it so much easier and for these actions, where we actually have these artifacts, whether it's like a provider archive or eventually for the otp release, where we're going to have the gzipped tar balls for your architecture and your operating system. What we do is we I'm sorry, I'm losing my stuff. We we attach that to the github release.

B

So if I look at the this is just like a test release to show you what this looks like. But when we release the wasn't cloud: http server provider, that's going to go into our azure container registry, but it's also going to be present here on the github release, and so that gives another option if you want to um not exactly sure the exact scenario where you like, wouldn't have wash or wouldn't want to download it from the azure container registry.

B

But it's also available here really easy to grab and this will kind of facilitate how we're going to distribute the otp release as well.

B

So I think that's most of what I wanted to talk about in terms of github actions. um Our otp actions are probably some of the most complex just because there's lots of steps for uh elixir for for building, and you know the the cargo clippy equivalent like credo or credo. I don't know and dialyzer for elixir, so I think there's some opportunities to move these into the common action repository and largely across the github org. We have ci with github actions in place to like publish our crates and publish our artifacts.

B

So um it's been a lot of work up front to get this all working well, but now you know once it's tried and true and we release using automation. We can just build off of that forever.

B

um Anybody have any questions or want me to go talk about any specific um step. Hopefully I didn't gloss over any any. You know cool things about the common actions or anything, but I'm pretty proud of how this turned out.

C

Yeah, that's pretty sweet um yeah, I didn't know. I didn't know that the common actions could be so simple. I thought that you I mean, I think I wrote or debugged at least an action that was written in typescript. um I didn't know that you could just do them so easily and yeah. It's pretty nice.

A

Yeah great tour brooks thank you so much. I really appreciate you taking the time to do that. um A real, quick uh david. I know um you uh have a bunch of stuff you want to talk about today. uh Steve you mind, giving us a quick update on um uh the 0.5 release of losing cloud. Yes,.

D

uh Sure and um uh just wanted to say, uh I think it's really awesome this. The uh automation that brooks is doing. um We have over 30, probably closer to 40, rust crates, um and so any code savings we can do uh is a big is a big win. Any avoiding duplication is a big win, uh so I'm going to share screen with the zenhub board.

B

D

So this is a release report.

D

uh All right, you're, seeing my browser that says, wasn't cloud 0.50 right, yep, um okay, this is a release report that shows a um sort of a burn up chart and we're still as an organization getting used to these things. um There are some stories that had a large number of points that we did in the end, so that this slope of our velocity is increased, um but we're tracking this we're tracking this towards uh completion, we're primarily working on documentation and examples.

D

um Here's a stuff stuff we're working on now includes uh the the github actions that brooks working on uh tutorials and demos um and so we're getting closer and closer. um I I put a screenshot of the of this graph in the slack chat yesterday. If people are really interested, I could do that more often, just let me know send a message in slack.

D

I could show you a couple of the couple of the some of the documentation.

D

That's in a pr.

D

How do you move this screen share demo, so you can access your browser tabs.

D

Zoom is blocking you from yeah. Zoom is blocking me um uh well, let's here um so here's here's one example of uh the run. We started start an actor, so this is what's going to show up on wisdomcloud.dev um walking through. How do you set up link definitions? How do you get started? We have uh you through the code. Generator there's also walk you through the update process. So if you update your code, then you need to push the result and update and test it. So all of that's in there as well.

D

uh So this is uh something is that's moving along, um should be in pretty good shape in the next couple days uh should be ready to publish.

D

So that's my update.

A

D

A

It's awesome, steve and I did try to print this off as like a pdf and export it, but it all like prints out goofy um uh stuff.

A

So we still, I think, have some more work to do on uh being more transparent on this or find a system that lets us accomplish the um organization of releases across a wide variety of repositories like what we're doing now, um so uh we're still uh maybe on the hunt for that um and uh screenshot to print yeah yeah, we'll uh we'll see how we can get that out um a little further here and I think, there's also um actually on the red badger side.

A

um Now that you guys you guys, are all added to the repo on um on the next topic here. uh So I think you guys actually would should be able to see this now. Is that right, steve like if you're in the org it shows up for you.

D

uh If the yes, if the, if they are members of the wasmcloud oregon github, you can view zenhub uh in with the zen hub free account.

A

Okay, uh well we'll see if we can uh maybe just circulate that link a little bit and see if it works for you um as we move in um david. I know. Oh sorry,.

D

Can I can I follow up on on liam's point uh more specifically. We we do want to be transparent about this, and uh I don't get too many responses to the weekly updates. um So uh if, if you feel like you have a way to describe the information that you'd like or a form, that would be helpful for you to follow along. uh Please, please. Let me know because I don't want to keep. I don't want to keep doing something. That's not useful.

D

I want to make it really useful and informative, so please send please send feedback.

A

Thanks, thank you. So much steve awesome work. I really appreciate all the facilitation you're doing and the organization there. I think it's super helpful um david. I think we're back to uh the discussion you guys were having uh when we uh when we uh I kicked off here, um uh I don't know, does it make sense for you or kevin, maybe to do a quick, high-level overview for what the discussion was about and then you guys can dive right back into it.

C

C

uh So people who don't know.

C

uh So the dream, the dream dream goal, I guess um is for you to say, um keep control apply. You put. This is my dream goal it might not align with anyone else's. um uh You put your application like a single resource that describes your application, that um something like kuvela can read um and also um some theoretical welcome cloud operator um uh can read um and then um uh kuvela will set up like pods like if you need to actually do the top level.

C

Like wasn't cloud um pods and like scale the wasn't cloud pods um horizontally, um then you would like you could get cubella to do that. um Maybe uh and then um and then question mark question what question mark and then profit.

C

No sorry so um and then uh so we feed like the intention to the last controller, um this theoretical license controller, um uh which sits like talking over gnats, um uh with, like a recording of what the intention is um and also the ability to query the hosts for what reality is, and the intention like the lattice controller's job is to make reality um like reflect the intentions that you have um and then uh and then I'm not actually sure, like I'm, not actually sure what this path looks like.

C

um But then I've never actually written a kubernetes operator. So I'm like the least qualified person to talk about this, which is why I'm picking um janitor's brain uh and desperately trying to depict anything, I think, has written one.

E

um I've got some ideas about what that back path looks like, um I guess, in situations where you would work where you're running wasn't on top of kubernetes or on top of multiple kubernetes clusters.

E

um I think it it's quite a nice idea to visualize like an argo cd um browser ui with um the tree of all the kubernetes resources and the status of those and- and you know, an operator will pick one of those up and potentially picks up a parent resource, creates child resources etc, which either do something or don't do something, and those child resources are often just used to store the status of those resources, whether they're, real or not, because at the end of the day, they're just records in etcd, I mean it's just just a custom resource so think about the way cert manager works.

E

A cert manager creates a custom resource for and ordering a certificate for, the challenge for the actual certificate itself, and you can cuddle describe any of those resources and see whether you know where the certificate um is in its issuing workflow.

E

um So I don't think it's uncommon for um an operator to create child custom resources and kubernetes, regardless of whether they refer to any physical thing or not, um and so well. Cooper is a is an operator for kubernetes that works with oam um and it I thought I'd look into it more because you know I don't particularly think we should necessarily spend a lot of effort writing an operator if this one already exists. It works with the file type that we're.

E

You know with this crd that we're um going to use- and I also think, there's some benefit in having an oam custom resource application, resource that contains potentially a mix of actors and capability providers for wasn't cloud and kubernetes, pods or deployments or whatever that what um serve microservices, that you know talk to each other like a redis database or an ingo skate, where, whatever whatever it is, so I think um something quite nice about having an oam resource that contains the whole description of your application.

E

Regardless of whether it's you know bits, the actors and the capability providers are working running in wasn't cloud everything else running in kubernetes, um and so, if we follow that model further, then the operator would also create custom resources for actors and capability providers that were just dummy placeholders effectively for the real things running, wasn't cloud, and then they become a receptacle for um status.

E

Information coming back from what's cloud so, for instance, if you were looking at your whole application as a tree in an argo cd web browser, you would see the actors and capabilities uh capability providers named and containing information which you can use, including the status like whether they're, healthy and not and they'd, be linked back to their parent, which would be the oem application resource.

E

So that could work quite nicely um so you'd have you could almost potentially have cuvela at the front doing the operator bit um producing and I've already got this working as a sort of like a stunt hack, uh creating application, sorry, creating actor and capability provider. It's just dummy ones doesn't do anything else. Just creates those resources, and then couple also has this new new new functionality called workflows, which you can use to sort of orchestrate how those resources are actually created. Maybe there's an order to them.

E

Maybe you want to do something at the end or at the beginning, or you know like um a bit a bit like helm hawks, I guess, but on steroids, and it looks really cool. So um you can use the q templating language to write these workflows and they can call out to do things, and so I actually got a cube bella watching an app application resource, creating access and capability custom resources and then once it had done, though, that it then posted the parent oam application resource to an http endpoint.

E

My online macbook, which I just used to demonstrate the principle. Basically um so it I guess what I'm saying is: cool vala could create the resources that just act as receptacles for status, and once it's done once it's done those um on every change to the oem. It would basically make sure that everything was in sync and post it somewhere and that could be pushing it to an outstanding topic, so we could actually have with cooper. We could have that goodness operator without doing very much at all, but it's just an idea.

F

So I'm curious, I'm curious to see how that works out, because when at least from my uh poking around with cabela the it's mostly like a templating engine, where you know uh you, you define what you want in terms of oam primitives and then you know, based on the type of thing that you've got loaded in cabela.

F

It's then able to produce text files, gamma files and things like that based on parsing through that oem file, and so it may be too generic uh in some sense, in that the the lattice controller is designed to operate on an entire oem, app spec model, and uh so whether or not cubella is involved, the lattice controller doesn't really care. Although all it's really using is oam. And uh so you know, people with more experience can certainly decide.

F

But I I wonder if cubella might actually be adding too much overhead to the the process when a thinner controller might might be the way to go, because if you're upset it depends on the granularity of the crds that you're monitoring right, you could create a crd for every actor and another crd for every capability provider, or you could create a single crd for the lattice controller's entire oem model and then just keep track of the status of the entire deployment.

F

As reported by the lattice controller, because the the plan there is that, in addition to the lattice controller, managing the the deployment uh on the lattice, it's also going to expose query functions that you can use to query the status of the deployment and potentially do things like get. You know an event stream so that you can hook into things like the change event on a crd and kubernetes.

E

I like that, I like that idea, even more actually, because um you know what's implied, has it has its has um the wash board, and you know its own um set of tools for reporting all the status of everything, um and you really don't want to necessarily replicate that in kubernetes. So I I like the idea of having a less granular model. So in other words, you wouldn't rep, you wouldn't model the wasn't cloud environment in kubernetes at all, yeah right.

F

Yes, that was the thing. Is I worried that attempting to create uh crds for uh that have parity with low level wisdom cloud primitives might.

G

F

Adding more work than you need yeah, because the lattice controller is already keeping track of all the low-level primitives and bubbling up high-level events, so it might be worth it to just create. You know a crd that represents a single wasm cloud deployment and you know, then all you have to do is ask the lattice controller, hey. What's the status of deployment with this name and and now your operator is complete.

D

F

H

Because for the status you can have and put whatever you actually want in there, because it's actually its own structure, as I'm sure, and everybody else is aware of um and yeah. If you just have one- and I guess so, it sounds really like the intent here is. The operator is just a way to report status back use, so you can view it through keep cuddle or whatever else, actually interfaces with cube cuddle or that api. Otherwise, it's actually it's basically a dumb listener right. You don't want any actions to go through.

F

Right, I mean, I think the only thing you'd want to do is keep cuddle apply, uh which would then you know downstream produce the lattice controller's oam file uh and send that to the lattice controller. And then you theoretically have a one-to-one correlation between the crd and the wireless controllers. Manage deployment.

G

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

H

C

H

C

Be responsible for stripping out anything that elastic control doesn't understand, or is it the lattice controller's responsibility to do that.

F

Well, lattice controller is essentially going to have an oam it. It is aware of its specific oam, so it knows that what a a lattice controller's app spec looks like, and it knows which traits it supports on the types of components that it has.

F

So if you try and send it components that it doesn't understand, it will likely reject the oam file.

C

Right yeah, so so in a model where you wanted to describe your entire app as a single im file, as in the theoretical feature, I guess you.

F

Would yeah so one of the one of the components of your larger app would be the wasm cloud oam model.

C

Right, okay, you wouldn't trying to interleave it with like, like a cabela, I am you'd, have you'd have a couple of oem for like the like infrastructure app. If you want, you want to call it like the framework and then in then then a separate swarm aside for the wasn't that I guess that makes sense.

E

C

E

Tricky, though, because because oem has um specific um groups and kinds right, api.

I

Version groups, kind of yeah.

F

The one for the wasm cloud app spec, uh the the group on it, is different, so you can use that to differentiate between, like, like say, a a super or a higher level, app spec and then within that you could have as one of those components maybe a wise and cloud aspect.

F

Yeah. I don't know how I don't know if the structure supports that kind of nesting, but um it might work.

E

But either way cooper would ignore the ones that didn't have the same api version, etc.

E

So they could be completely independent rather than nested. Obviously,.

C

Yeah yeah, okay,.

F

Yeah you, you wanna, you wanna, set the the kubernetes controller up so that um it works just on stand on. uh You know, standard operator, primitives and wouldn't require um interrupt with cabela.

G

F

Because, if cabela's doing its job, then it can pretend to uh it's: it's essentially just a source of additional text files right, so it can it can shove, yaml files or whatever kind of files you want into any target. You want.

H

Yeah yeah, so it sounds like I mean there is a demo coming up on friday or well. Actually, sorry, you guys have your recordings, but for the lattice controller it is theoretical at this point can do we have an api spec for it at this point in time where we can design, I suppose the.

G

Crds the api for the lattice controller is under.

F

Construction, the there is a there is a spec for the oam file or the oam um object uh that it accepts for uh for putting a a new management cloud deployment. So uh that part is at least done. So you know creating a new deployment should be fairly straightforward. The other stuff like querying the status and things like that, those those data structures will have to discover as stuff gets built, but uh I'll post straight to the wisem cloud, slack channel uh a copy of the the latest oem file or a sample oam file.

F

So you can see what it looks like.

A

A

uh It's a really robust discussion um uh kevin. I know uh uh if you're sharing those uh do we need to, and I think I really appreciate- we've actually learned a lot by trying some quick pocs around this. um Does it make sense for us to uh put our collective heads together and capture this as a um as an rfc or an rfp? Maybe the lattice controller started out as an rfc. It's already there absolutely, um uh but to connect the rest of the stack up here. I think that there is some pretty significant alignment.

A

Do we do a operator rfc um so that we get a line there and then we can maybe plan. You know execution yeah. That sounds like a good idea. um uh I not sure that I could capture the rfc here. Did somebody feel like they were following along? Could um get this sketched out in a way that in a into a ticket up on github, so we can um uh get this rfc crafted? Did anybody want to take the lead on that.

E

All right, we can.

C

Flinching you david you're, looking at me, okay! Well, um uh if it's happening it's happening next week, okay, yeah.

A

C

A

Yeah, this is important, not urgent. uh uh Thank you david. I really appreciate that um that'd be awesome, I'm quite happy to help with it as well, david just.

H

Feel free to ping me on pacific time us but yeah, okay,.

C

C

C

Okay, we'll work out like so in terms, I feel, like we've got in our heads like a reasonable idea of like what the minimal like operator actually looks like in terms of its traffic flow um in both directions. um That sounds good um and you've got a skeleton.

C

um Go implementation of like a generic operator like that, probably will fit reasonably well into the model, and I guess the the nats communication go, has reasonably good nats libraries, because nasa's written and go yeah.

F

Go has the uh the first class they they get all the goodies before everybody else does for nasa.

C

If we jump, if we're jumping onto um like implementation, not just specification um like an implementation, is based off that um that uh go based like skeleton like it probably is a reasonable start. um So yeah.

C

Okay um right we're writing things up, um but yeah like let's get demo working for coupon um this week and all right I'll finish it. Unless someone else jumps in and starts a pr like we make vr for the rfc and like have multiple people make sub pr's to modify it. If you want um yeah.

E

C

um Yeah, if no one else does it I'll do it next week, all right.

A

Absolutely brilliant! um Well, thank you, everyone for that. uh The discussion I you know, um I know not.

A

Everybody um feels positively about kubernetes, so I just wanted to maybe socialize my perspective on it and see if there was some alignment you know I have really advocated for us um investing here and building tooling and um connecting to where people are today, because I want to be able to take them to a different place and if you in big enterprises, my perspective is that it takes so long for them to build up the infrastructure and get all their boxes, checked and aligned with policy and procedure and have security and scanning and governance and legal and compliance and all the other orgs um there that we can dramatically accelerate the adoption of wasn't cloud by um investing a little bit of time now in order to be compatible with their community and the ecosystem.

A

And obviously my hope would be is: is that as we mature, our ecosystem and other options emerge or other design patterns and other architecture evolves that we could uh bridge them out there? um Does anybody have a different perspective that they want to share or feel strongly against? That, or for that.

F

You know, I think that just generally goes along with the strategy that we've always had, which is to be compatible with all of the other things, but not.

G

F

Upon right- and so that's essentially the reason for the the lattice controller architecture that we're heading for which is it'll, make it very easy to build a kubernetes controller or kubernetes operator, but that doesn't preclude any other framework from sitting on top of the lattice controller as well right yeah, and that was.

A

F

Sorry kevin go ahead. Yeah, you know you can deploy the wasm cloud host in kubernetes pods just as easily as you can on uh bare metal and on raspberry pi's and so on so they're the we continue to have, as we you know, add features. Our roadmap essentially requires that you know we. We maintain a a firm stance that our stuff is compatible with, but not dependent upon.

A

Yeah I mean this: is you know the world um the world kind of changes? You know, as we kind of think about these. um You know epics of computing here um you know with a web assembly, because the fundamental nature of the unit of compute is kind of changing. Here right, you know we're not moving from a abstraction of a virtual machine, an abstraction of a linux environment, we're now to we're back to an abstraction of a cpu. You know it's almost like recursive when you think about it.

A

That way, except there's, you know no operating system in here. We've just got our wasn't cloud shim um that pulls us up into the future in the future. We will have. You know, wazi um sitting on top of this stuff, um to give us some. uh You know some operating system type primitives, but the way that we've always built wasmcloud is so that it's completely compatible with and those features would just uh flow right into um either a capability provider or just naturally into the project there. But that's uh that's um where I've done this.

A

I had a couple of people. Ask me to do a version of this, but to put the orchestrator technologies in here so to you know, put a cloud in the stack. You know to put um a kubernetes in the stack here, so I may take another pass at these.

A

If folks have thoughts on um trying to figure out how we can surface those, but again, this is the world that I feel like we're really building for um with a huge emphasis down at this end, where I mean we run everywhere, webassembly does, and that includes you know, native silicon without kubernetes. Maybe I should illustrate that on here, pull kubernetes back a bit on kubernetes directly in the web browser or directly on the devices themselves with the sort of cross connections there strategically.

A

Does everybody feel aligned that this is the world that we need to build for.

E

I can imagine lots of hybrid applications running in the cloud where parts of it are implemented in muslim club, but the remainder is still in kubernetes or something and you just gradually gradually move it over.

A

Yep yep, I I mean the the investment that the the years it takes for an enterprise to support this, which is what they've spent the last seven years building is not easily turned over. You know it involves. You know um in large, highly regulated, orgs policies and procedures going to the board of directors being approved. You know and then coming back down. You know at least the um the policies and then the procedures you know, like the actual. How do we do this?

A

You know getting implemented here, so it's going to take the enterprise um a while to catch up with you know in a world where you know you execute in other people's environments.

A

um How do you think about writing policy and procedure for that and they're already doing it now with stuff like web browsers, but it just becomes more real to them when they're thinking about executing on like an alexa or refrigerator, or you know something down at this end where we may not have kubernetes, you know, there's some lower bound where we stop kubernetes or we or we don't run this in the browser um so I'll.

A

um I think it's just really helpful for us all to kind of understand that um why we're you know doing these investments in and being compatible with, where the enterprise is today, uh so that we all make sense uh strategically there um open floor. Any other comments on that before we just move on to another topic.

A

All right super, a quick community uh updates. uh We've got the watsoncloud.50 release party being planned right now, so we'll do a try to do an evening with suburban and a one-hour demo sponsored by cosmonic. um We have a wasm day coming up on october. 12Th schedule was been published, it's about to be rearranged, and we've done a draft of that. We'll publish that next week, where we call out who's remote and virtual, and then who is in theory in person right now.

A

I still am skeptical personally that that's going to happen, but I do plan on being there um and uh uh out at the booth and then, of course, we've got kubecon uh eu uh there's another community event that was announced uh on september. The 22nd uh wasn't three something or other I'll link it in the notes. um It's a walk through and make sure we put it on slack in case anybody wants to attend. uh There will be a discount coupon put in this week for anyone that still needs to register for kubecon or wasmday.

A

I will likely have some additional free passes this year. If you're waiting to the last minute, you can ask me for those. If I have any left, uh but we're gonna put a 20 discount code in uh through the wasm weekly newsletter this week and I'll drop it into our slack as well in case anybody wants to grab it any other community news or activity folks have anything going on. They'd want to highlight.

A

All right open floor, any rfc rfcs to circulate- or I think we're all uh you know, hustling on our kubecon um deliverables and the 50 release anything else.

A

All right well incredible, meeting today: everybody thank you, so much I'll go and stop recording and we can hang out as usual, cheers.