wasmCloud Community, 10 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 20-Dec-9 wasmCloud Community Call

Description

wasmCloud is a dynamic, elastically scalable WebAssembly host runtime for securely connecting actors and capability providers.

https://wasmCloud.com coming soon!
https://wascc.dev

A

And the cleaners just turned on the vacuum: isn't that wonderful? Isn't that always how it goes.

B

Strangely, I don't have access to the dock, don't know why.

A

I'll fix that make sure it's on it. Okay, so uh welcome to december 12 uh 2020 wasn't clock community meeting, uh so we've got two new members. I think uh only one dialing in today, uh jordan. You want to take a quick minute and introduce.

C

Yourself, oh yeah, sorry caught me off guard hi, jordan rash. I am um here more or less to learn and see what this community is doing because of uh everything. I've read it's exciting, um I'm mostly a security analyst background, but uh you know we're all here to learn new things. So, thanks for having me, I know most of you hi bill.

D

Hey jordan, nice to see you.

A

Awesome and uh we've got someone else joined uh warren um who's. Gonna try to make the call maybe run a little bit late. We gave him a chance to introduce himself um a little bit later and uh we're gonna we're gonna start a new format based on some feedback from last time, we're to try to start each meeting with a great demo.

A

Well, let me not qualify that with a demo doesn't have to be a great demo, because I want to set expectations too high. uh So why don't we turn it on over to brooks who's going to talk about um the uh new uh repple uh functionality, and let me give you uh presenter brooks.

E

A

Floor is yours,.

E

All right, uh you all, should be able to see my little uh terminal screen here. Let me know if you can't um so today, I'm gonna be showing off a couple more things for the uh wash shell, the wasm cloud, shell. As you can see, we have a beautiful new ascii banner here for the new rebranding, so marvel at that, as you will now I've. I've shown off this shell a little bit before, just to give a brief overview.

E

What this show was created for is to consolidate all the command line, tools that we had for wash uh or for wasmcloud into a single binary so that it's a lot simpler and a lot easier to get started with wasmcloud and to develop in the long term.

E

So originally the first things that we consolidated into this were keys and claims which these two sub commands were used for, creating signing keys and signed jwts for securely signing actors and capability providers.

E

We now have the par sub command for packaging and signing capability providers in an efficient compression format. Two new commands that I can show you today are the cuddle command, which it is pronounced that way, and it is how you print you interact with a wasm cloud control interface, which is a new fancy way of saying that we're interacting with the wasm cloud lattice and the main subject of this demo today is the up command, which is an interactive reply for doing wasm cloud development.

E

That really makes it easier to quickly uh iterate on on what you're doing, while you're and get immediate feedback for when you're developing with wasm cloud. So I'll go ahead and get started and show you what this rubble looks like your right. When you launch this reply, you're greeted with a couple of different panes here, so the main ones up at the top are the reppl which takes in your input and any command output shows on the right and then on the bottom.

E

We have a really uh powerful logging widget that allows you to view all of the different levels of logs that are coming out of your programs, while you're developing, and you can also adjust the verbosity of these logs. If something is being too noisy or you want to really hone in on one thing, for example, you can see a couple trace messages from that subscriber we can drop that down to debug to hide those overly verbose messages.

E

This is really nice, because, when you're doing wasm cloud development, there's going to be messages coming from all over the place from the host from your actors, providers, etc. So it's really nice to be able to manually control this to. However verbosity you want now just to show you some of the things that this repo is set up to do, we can type in help to get some nice uh nice helpful um help text here for the different sub commands in the shell.

E

What I really want to show you today is how easy it is to start and stop capability providers and link them together so, and we can also gather some more information about what's running in the lattice currently so to get started. We can see what hosts are currently running.

E

We can see that we have one host which was launched automatically when the rep will launch so that we can do some local development.

E

We can take a look at the inventory of this host, which is pretty empty as we'd expect it by default. This host comes with a couple of different labels for what kind of operating system architecture is running on. I added an additional one called repo mode.

E

What these are for are when you schedule your actors and providers, you can specify what hosts they want to be scheduled on by running them through an auction and having them be assigned to assigned to a host very similar to when you schedule a pod in kubernetes, and you have taints and tolerations to decide where it goes so to go ahead and get started. I want to show you what it looks like to run an actor on this host using this rebel.

E

So the start command is how you start any actor or provider. You can also take a look at all of these commands. They have additional help text for, if you forget what you're putting in here, but to start an actor here, we just simply have to give it either a path to a local actor on disk or an oci compliant registry link for an actor. So I have one already pushed up to the azure container registry that we have called kb counter, and I have two options here.

E

I can either directly schedule it on this host or, if I simply just issue this command, the start actor command is going to run an auction for where this actor would be best placed the first host that I get back from that which will be my only host since I had provided. No constraints is where it gets scheduled.

E

So that's it. You can see in our output that we say which out, which, which actor has been started. What the actors, public key, is and there's additional information down in the logs. You can take a look at too.

E

So if we take a look at the host inventory again, we can see we have one running actor in here. The kb counteractor so being able to immediately issue this command and it gets scheduled on a local host is super powerful and much easier than I remember having to.

E

As far as like a quick start, you could run this anywhere, there's nothing local that I'm using other than the wash binary. So this is this is much much easier and the process for starting a provider is largely the same. So I can also use an oci compliant artifact here, like our redis provider, for example, and I can schedule that directly onto this host, the providers are a little bit bigger.

E

So I've supplied this flag just to make sure we have plenty of time which we didn't need to use, and you can do the same thing uh over and over uh to schedule. Things onto this host.

E

So I'll also do the same thing for an http server provider go ahead and schedule that onto this host and then after that finishes up, we can also check out the host inventory and look at our new shiny providers that have been scheduled there.

E

Now. The next part of doing wasm cloud development locally is going to link these. You need to link the actor to the provider so that they can communicate with each other. So that's as simple as taking the actor public key that we've gotten back after we schedule it specifying the uh the capability contract that we connect them with, which is just how the act knows what uh protocol it's communicating with the the provider. Here, I think I heard.

F

Somebody come around contract.

E

F

Kevin sorry, you need an underscore in between key and value for that contract. Id thanks.

E

And then we can also provide any additional configuration that we need like for this key value for talking to redis. We need a local redis url. So I believe that is just this.

E

Simple, as that we've linked this actor and this provider together. Now we can do the same thing, which is going to be to the http server.

E

One thing that we're actively working on is oops, oh god. What did I forget.

E

Address already in.

F

E

I might have uh there we go. I think I might have just put in an extra character somewhere.

E

So one thing we're working on instead of me having to go around and copy the actor public key and the provider public key for these links. We will be able to link to them by their oci reference, so the azure cr dot, io, slash redis, for example, and if we take a look at the host inventory here, nothing changes, for example, but where our links are present there, so that the actor and the provider can talk to each other.

E

Now, I'm going to try something we will see if the demo gods have decided to stop smiting me for today, because immediately before this demo, I tried to curl this actor and it didn't work.

E

I figure it wouldn't end up going through, but I had to try for the for the live demo, um so it looks like there's going to be more that I can show you next week for the full developer experience, but this really simplifies what we were doing earlier with I mean it, it may be easy to start up a project by writing it in rust, but being able to have this immediate feedback in the ripple is is huge and anybody can do this.

E

I don't have anything like I said I don't have anything on my computer. That is, you know these capability providers or actors. It's all pulled from a remote source, so we can really easily get people to run our examples as well yeah. We sort of imagine.

F

That this rebel will be a great tool for getting people introduced to wisdom, cloud and dealing with actors and providers and how it all works. But once you start building actors or building providers, we figure. This will probably be an invaluable tool to just have open all the time in another terminal so that you can poke and prod the stuff that you're building.

A

Brooks I thought that was an incredible demo. Thank you so much uh ralph. I know you got a reveal. uh Well real, quick. I guess any questions from brooks or anybody who want a discussion or feedback or comments form.

B

F

Yeah, it's amazing. I can't wait to start using it all day, long.

A

Awesome well, let's keep the feedback up and uh any anything any features you guys would like or anything in there. You know we want to build stuff that people want to use and um you have an evidence. You're a road map, a ralph. I think you need to peel, so you want to maybe give it uh over to you real fast and then we can. Let you get out.

B

Sure, just real quick, uh we'll postpone chris, you know working on the docks for um critical stack and crestlet and so forth uh um for this coming week. So we'll make sure we have that done, uh and I wanted to uh call out the blog list that we had discussed last last week. We've got a hack md there. Everybody should be able to add things, so you can test it and see if it actually works um the. If you refresh that, I think we just published uh we've added.

B

Let me see if it works sure enough. The post host for the first one is the real link to the blog, because we just published that so you get the idea, we're just basically going to add these. um As you see fit, uh taylor here is going to publish a one, that's more on rust and not quite as much on rasm that'll be in there, but we're just gonna sort of fill out and anybody can post anything all the interest.

B

I'm gonna sort of start, adding the ideas that we have over the next coming two three months to this, and so just free form. I'm not expecting there's any problem other than if I screwed up some of the permissions or something which is entirely possible. So I just wanted to call that out, make it available to everybody and uh power on, and uh I gotta.

A

Run, thank you. So much ralph have a wonderful thanks: cheers: uh okay, uh coming back up to the top um here, uh just a quick note about wasm time uh the 12 or the uh thanksgiving day meeting uh was uh cancelled, of course. So the next meeting is tomorrow at one o'clock. um If anybody wants to get loot up on that, just ping me and I'll, add you uh there's a meeting agenda for that already posted uh and um uh there's a pull request in place for the rfc.

A

It's basically the rfc for the rfcs, the meta rfc um I'll. Give you guys an update next week, but there's a number of things listed on the agenda that are interesting to us around uh wasm wazi nn, which is the intel, contributed, machine, learning, stuff, uh uh potentially being merged in and uh and a few other things.

A

Those are um no different than what we updated on last week and then I real quick wanted to um see if one of our uh folks here from the team would want to talk about this, introducing yo wasm post that I saw get dropped today. Anybody want to speak to this on the call.

G

Again, if nobody else does um so so one of our engineers, who's in new zealand and consequently, you could never make. This call has been working on a framework for generating for scaffolding out webassembly projects very very easily in a way that integrates with visual studio code and and provides support for a number of different languages. So ivan's been working on this as his sort of research friday project for the last several weeks or so um and got it to the point where we felt pretty uh comfortable releasing it.

G

It currently has support for c rust and assembly script, we're working on adding support for swift. um Now that swiftwasey is working pretty well and it's kind of a nice little plug plug-in based generator very easy to modify. um We he he designed it so that we could um continue to add new features as we go and again, the the point was to basically uh make it as trivial as possible, given the state of the tool chain for a developer, to set up a webassembly, compiler and and environment for their particular uh language.

G

So I think, by the end um ivan had it integrated with the vs code debugger uh and had the default set of um vs code uh actions. I I never can remember what those things are called for uh for compiling and things like that um and had support for a couple different frameworks, including last I checked. It was just wask and wazzy, but I think he might have had vanilla uh wasm with neither of those enabled.

G

So it's pretty cool, I'm really glad that he released that and wrote up a pretty good blog post about uh about how that works. On top.

H

Of that, I think, there's also support for github actions that automatically push to container registries that support webassembly modules. So for people who do push modules or wasn't cloud modules to webassembly to container registries, it might be of interest as well.

A

uh Very cool stuff: um is there any time here to you know what phil's doing with his wpc? You know generator stuff you know is there are there is, I feel like there's a little bit of overlap between uh between some of the work there. Is there any um any potential to like collaborate on that.

G

Yeah, I think how it would work, because uh the the yeoman generator would just call out to the external generator to do that. Work. That's kind of the way the yeoman world typically works, so it should be pretty easy to integrate the two. If we want to do that.

H

There could be a plug-in specifically for generating wpc compliant modules.

A

Yeah, that's very cool yeah. Well, I wonder uh kevin. Maybe we'll take this offline. If there's, um if maybe we should um look at, you know leveraging yeoman, because we get all the other stuff for free if we get all the vs code and all the other things we need to like really just you know think about how that can help.

F

Yeah, the trick is to figure out which, which tool is the the parent and which one is the child, because wapc has its own scaffold, generator that it uses for code generation um but yeah. It's it's definitely worth looking into I'd like to have some form of uniform way of creating all these different libraries.

A

I will just throw another log on the backlog uh and we'll see if that one ever makes it to the fire uh uh to burn down. uh Let me see here, uh wpc generator, okay, awesome, um so I think right now we can go into oh. We got some to do review here. Share my screen a couple of just run right through these.

A

I think radu and brooks you guys had a did. You guys have a chance to look at that um that pull request that uh you guys were discussed last week.

H

I think there was another round of reviews, I'm not sure if it was merged just yet.

A

Okay, anything um anything we can all help with or any do you need any support or anything.

H

uh Not that I know of I think and I'll. I should probably let taylor talk about this some more, but I think it's in a state where we could look at actually merging it without committing to anything right now.

A

I mean I was leaving this up to you raju, so if it's good with you, we might as well emerge it.

A

E

Now I'm gonna.

A

Check the magic 8 ball first, you know is that is that, on the on the agenda I got. I think I got like a 30 sided die over here we can roll for a crit. You know uh if, if you're, the thing is what I do is if, since you know more from the lci side, if you are okay with it, I am okay with it. um I think I did tell you why the tests were failing, brooks in some chat somewhere, there's three crestlet channels.

A

I have to check internal uh kubernetes and wasn't cloud so um in one of those places. I think I mentioned uh what the pot, or maybe it was on the github issue. Anyway, I mentioned it somewhere about what I think your issue with the test was, but as soon as those tests pass as long as writer's, okay with it redu can go ahead and hit the merge button. I'm fine with that.

E

Okay, I might need to double check on um what that was. I know there was an issue when I first submitted the the pr that, like some runner, was out of date or something like that, um we fixed that. I think, if you, if you rebase, that that particular.

H

Issue is going to be fixed.

E

Okay, then, I will take a look at that and then touch base with either taylor. You rather I'll put it in the crosslit channel in the wasmcloud slack.

A

All the things all right, thanks everybody, uh and um so the links by the way for um ralph's blog post list are here uh and uh linked on the agenda, also tweet, those out from uh from all the accounts just to try to get some attention on those. um I guess we're now down back to the dot 15 review. I don't know if anybody wants to start uh brooks is out for the rest of the year on holiday. um I'm sorry, chris is out for the rest of year on holiday, but uh kevin uh bill.

A

um Do you guys want to get anybody want to start.

F

uh I can give just a quick recap: um we're still basically, basically just finishing up the the last few touches on the control interface stuff um right now. We've got auctions being tested and we're working on those and um the live update. Stuff is now in we're just uh I just need to set up some docker scripts so that I can make sure that that all the live update, stuff gets tested properly, uh but other than that it's it's uh that's all good to go.

F

There's a couple of distributed systems, type issues that we need to take care of, but um none of those look uh look insurmountable either. So it's all going well.

A

Are there any in particular, you want to like to just speak to as an overview.

F

I get kind of the high level distributed systems issue is when, when you're running as a lattice there's some information that the hosts are sharing with each other and some that that they aren't and that's currently creating a problem and producing some very interesting edge cases.

F

So, right now, the lattice will share its claims cache and it shares the link definitions, but it doesn't share the mapping between the oci reference and the public keys of the things that we've retrieved from the registries so, depending on the order in which you launch things and where you launch them from the lattice could artificially tell you that it doesn't know what you're trying to work with, because the oci mapping is stale. So there's two issues for that.

F

One is the specific one to deal with replicating the oci references and then the other is we sort of need a more generalized, more robust way of dealing with distributed cache data right now? It's it's ad hoc and they're specific events that we use for replicating very specific pieces of data, and what we really need is probably another actor inside the wasmcloud host that deals with a distributed cache and per our requirements for lattice that distributed cache system cannot be central point of failure.

F

It cannot use single server and it needs to work in partially disconnected mode.

F

So there's the the short term fix, which is we're just going to manually, replicate the oci image mappings and then, as soon as that's done, we'll start looking into a more robust way of dealing with the distributed cache.

F

I'm sure everybody's familiar with it. But this problem is something that virtually every system has kubernetes deals deals with it with things like uh fcd or console, and even though those things are clusterable, they are single points of failure and that's intolerable for the lattice requirements.

A

um You know evan baker had written uh e2d, which was uh taking ncd and trying to solve some of the management life cycle on it. You know, I don't know if there's some lessons learned on how to make the kubernetes workflow simpler um that we could, um you know we could learn from. uh We can learn from there. Yeah.

F

At a at a really simple level, uh most of the the information that we're that we're trying to distribute into cache could be represented as ad-only lists as a crdt. So we don't necessarily need super complex, but um you know: there's some spike work to be done and some research to be done there. How.

A

Do those lists get partitioned kevin? Is there is the idea that all the you know all the things know about all the things so the way crdts.

F

Work is they are um perspective, based uh diffs.

F

So what you replicate is a context and then a value relative to that context, and so, if one host adds an oci image map, then it adds an oci image map to the context as seen by that host at that time, and then it would replicate that list edition and the context to the entire rest of the cluster and then, as each other item in the cluster, receives both the context and the new value. It's able to resolve that in a mathematically provable way that there will be no merge, conflicts.

A

Is there a risk of you know? Let's say it's a giant deployment. Let's say it's, you know a thousand or ten thousand. You know um you know, actors you know running here. Is there a risk of having you know like a list? That's just gigantic. You know like for you know for when you think about like the peer-to-peer context here, I'm imagining sort of like a hub and spoke model in my mind here with some central capabilities and then a lot of distributed capabilities.

A

Let's think like, um like the uh uh the avsb demo, that we kind of like toyed around with what, if we were there were a million adsb sensors.

A

The every sensor certainly wouldn't want to know about all the other peers. Would it.

F

Yeah, so some of that is future problems that we we don't really have right now I mean one of the a really great problem to have. Is somebody in production running so many wisely cloud hosts that we need to solve for that problem right? That would be. That would be fantastic, but um you know by and large the crdt approach. uh Those are diffs, so the impact on the wire is minimal.

F

The what what you end up with.

F

Is the in-memory impact of having those huge caches, and so what you would essentially need is to have each of the the hosts be aware of how much information it needs to cache, so uh hosts that are running on an arduino would would have a different caching strategy than the the full back-end cloud host and so on.

F

But again, this is all conjecture at this point. We need to uh to do some research and spike it out and build some prototypes and see how that works. Okay, super, but the the core idea is right. Now we have two or three different manually replicated bits of data and, as we start, adding more features, we're going to need to replicate more data in a more reliable way and so the the existing ad hoc mechanism. This just is not sustainable for scale.

A

um Can I call attention to two of the rfcs that you kind of posted this week um to maybe highlight that we would love to get some feedback from folks on the call around what we're trying to uh like what the vision is here uh I would start with. Maybe the web browser uh one yeah.

F

So the the idea is, we have so right now we have essentially one runtime, which is the the rust base cloud runtime uh and we're planning on building multiple different specialized or fit for purpose runtime. So there'll be one so right now you can compile the rust runtime and target things like raspberry, pi's and other small constrained devices, but you can't do things like um arduinos or tiny edge devices for specific host host environments. We need fit for purpose, runtimes and so they're.

F

The two that were that we're thinking about are the web browser itself and swift for ios and mac devices, and so we have rfcs open for ideas on what those implementations might look like. But, more importantly, I I wanted to make sure that we captured the things that each of these host runtimes would need to be able to do at a bare minimum. So they need to be able to connect to the lattice. They need to be able to do basic multi-threading.

F

They need to be able to verify at 25519 signatures and um beyond that, it's it's open for comments. In terms of you know what people have as far as ideas go for implementations and designs and how that stuff will work.

A

Okay, thank you so much anybody have questions for kevin.

A

Super billy up next.

D

Hey y'all, uh so this week we got a couple: pull requests merged for packaging and shipping out wash and wasm cloud as debian and rpm packages. uh So currently there is amd64 um support out there. We have arm support built into our github action that will get triggered on release.

D

uh It leaves a lot to be desired right now, uh since uh github actions doesn't support arm runners um but yeah, it's a good start, um ignore the wasm cloud version, sorry kevin. That was just kind of a test. uh It says 1-0. Obviously, that's not the case, um but yeah yeah everything's um moving forward pretty well.

A

Okay, that's great, I'm gonna ask everybody to hold on just one second, because they're right outside my door, he's gonna, ask him same thing.

I

F

A

All week they are wonderful, I'm so grateful that they still come, because I do not have time for that kind of stuff, but I'm gonna have to do that part of the house, not myself. First of all problems I'll, stop complaining. uh Okay, uh so I think uh brooks. Did you have anything else you wanted to run down or was that all.

E

Yeah, that was all.

A

God bless it now, sneeze, it's getting worse. um I guess that opened it up. um I don't know that we have anything else so just kind of skimming down the um uh the to-do list. I guess I can go uh pretty fast. I think people are seeing some of the transition across repositories and web pages as everything switches into awesome cloud working on the new site and trying to get it launched as soon as it can right now everything just forwards over.

A

We also have gotten some basic. uh You know some additional metrics and monitoring up for, like um you know, twitter and trying to just continue to drive like you know, announcements and features um and um uh logo uh is in flight. uh When it comes back, I'm gonna get some nice hoodies ordered for everybody uh and t-shirts and stuff like that and we'll get some schwag. You know um spread out as we kind of come up. um You know on the on the dot, 15 release and start looking ahead.

A

um Blossom cloud is a real thing. Now wasn't cloud llc uh I have talked to. I had a follow up with till recently on how are things going with the foundation uh because, again, the ultimate goal is to you know, get the collection of you know, keep this open source and protect it. um We had uh what I wanted to make sure that we were doing, though, was being good stewards, so I've registered in an llc to hold. You know the trademark for wasn't cloud logo.

A

All of the domain names we own, like a dozen domain names related to wasn't cloud all of the accounts, uh so it's all in good order so that when the time comes whatever this is a year from now, who knows uh we're? Basically, we have all of our ducks in a row and it should, you know, make getting into a foundation much easier because we have a clear chain of you, know, ownership and and all of those kind of things. uh But you know just to you know double down. There's no!

A

You know um uh you know like it's a holding. You know entities so that we have all of this together and in one place any questions or thoughts or suggestions about that.

A

Okay, well, I think that is it there's no new adls to open up. I think we scanned issues. uh There's one pull request hanging out kevin and anything there to discuss, or is that yours waiting.

F

On feedback, it's not really waiting on feedback, that's just the one where I need to do some docker scripting to set up a local oci registry to get some integration tests working. Okay,.

A

Awesome uh jared: I still owe you a follow-up on like getting together on like um uh setting up like docs. You know a lot of stuff to work on together. uh That's what was on the to-do list. um I skipped over earlier, because I didn't do it yet and didn't want to call attention to my failure, but we can highlight that.

A

uh Definitely um anyone else you want to jump in anybody got anything. I want to share radio taylor.

H

I think last week we were you mentioned that you wanted to discuss about the networking and sockets api in wazi, I'm open to do that. If anyone has specific questions about it,.

A

Do you want, do you want to present anything or do you want floor, or I would love that if you want to just like sketch or just speak to it, even I.

H

Think the the easiest thing for people interested is I'll. Just I just shared my the write up that I did for it, but, uh in short, uh the uh there is a proposal in the wazzy community that has been kind of stale for the last couple of months, but in short, uh that proposal uh aims to add, essentially full berkeley, sockets support into the top-level wazzy api and the the article that I just linked uh tries to do a uh essentially a shortened down version of that with just that.

H

Just starts sockets in in wasm time and is able to essentially start client sockets from rust and assembly script, with example, implementations on how to implement the networking api from both uh in the rust standard library and a brand new one in assembly script. So uh it's not suited for server side applications. So if you want to start us an http server in you know wasm time host, it's probably still not a good idea, mainly because of the lack of uh multi-threading and other things, but for client uh connections.

H

It seems to be the the direction that was he tries to take uh for the record. The implementation presented here is by no means should no be, should not be used in production. It's just an example of how to do things, but it's fairly in in the same direction as the official proposal, which has received fairly good feedback. So I'm more than happy to chat specifically about implementation and other things with people interested in it.

H

It's just at the state of an experiment right now, but hopefully the uh the wasa community moves forward with the uh with the current proposal.

A

Yeah I've got a question I haven't looked at any of this, yet you know when you is there the whole. You know like baked into this. You know this uh poc work and I, and hopefully that is, doesn't come off the offensive to use the word poc work, but into this uh into this work is: do I what I still need to you know grant cape you know, grant these capabilities.

A

You know to allow this. You know behavior in this. You know branch here or is that you know work to be done yet this.

H

Is the main reason why this is and poc is the right word for it? It's just an experiment. The main reason why this is still a poc is that um the capabilities are not hooked up into the into wasm times: option capabilities this.

H

Just if you choose to fork this branch, it will grant a a tcp socket to any webassembly module that tries to open one so that that's the main reason why I'm saying this should not be used in production, but the uh the proposal actually hooks in the capabilities actually adds new capabilities to webassembly modules and defines them and uh in theory they should be plugged into azam time all up.

A

um I don't know what demo we have queued up for next week, but as you've already got some, you know poc and some test code here. Would it be worthwhile doing a like a walkthrough or demo next week? I should address it in that yeah yeah I'd love to see that yeah, maybe, like you, already have all your work here and you're already running the fourth branch, and maybe you could do a demo we'd have it on video and then we could. You know like share it with uh share with everybody next week.

A

um uh If that would be great I'd, love that uh who else does anyone else have anything they want to share open floor.

A

All right going once going twice, I'm just going to stop recording folks can still stand on the call, though thank you. Everybody have a great week.