Fluence Labs Community, 25 Nov 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Developer Community Call #6 | Fluence for JS developers

Description

00:00 Preparation
03:15 Intro
04:00 Updates from the core team
11:05 Self Id project demo, Fan Shuangying
25:18 Description of the Fluence
26:50 What are peers
28:10 P2P communication and particles
29:00 JS Peer implementation, where it goes into Fluence tech stack
29:37 JS Peer specifics, differences to reference peer implementation

JavaScript\TypeScript as a target for Aqua programming language
31:54 Services
33:13 Functions
35:12 Data type conversion

36:02 fluence-js package overview: Fluence Peer, Default peer, Fluence facade, Aqua compiler support
38:50 Fluence-JS : Fluence Peer Public API
40:52 Fluence-JS : Fluence Peer Internal API
41:38 Fluence-JS : Compiler support(1)
42:32 Fluence-JS : Compiler support(2)
43:08 Building blocks for a frontend fluence application
45:02 Part1. Installing dependencies
46:10 Part2. Setting Aqua Compiler
46:50 Part 3. Aqua code
47:41 Part 4. Integrating into typescript code
49:47 Demo time
52:23 Q&A

Speaker: Pavel Murygin
Host: Bernhard Borges

Useful links

https://fluence.network/
https://twitter.com/fluence_project
https://github.com/fluencelabs/aqua
https://github.com/fluencelabs/aquavm
https://github.com/fluencelabs/marine/tree/master/examples/greeting
https://fluencenetwork.notion.site/fluencenetwork/Fluence-Developer-Community-Calls-2733beddb3ff4e4986c4fcc830fc796d
https://github.com/fluencelabs/marine-rs-sdk
https://t.me/fluencedev
https://discord.com/invite/5qSnPZKh7u

A

While we're waiting, how are you fans what time is it for you.

A

I probably mispronounced your name.

B

I'll thank him in discord. Maybe his.

C

A

We have on so far hi james.

D

Hi, how are you doing bernard.

A

Good, how are you, how are things going.

B

Oh, I can't complain.

E

It's all good, that's good.

A

Doing things almost there almost there, it's all good.

C

A

All right, which is waiting for evgeny and dmitry, I think then we can kick it off.

A

You gotta rework that uh that uh one link on a the registration link, apparently.

B

Comment on that, okay, it's super annoying.

A

B

I don't know, maybe we just have to do it in discord next time.

A

Well, we should that live platform seems pretty cool. I haven't tried it so.

B

A

The the live platform they're using I mean I've attended a few in discord. Those live events pretty good.

A

A

A

All right, okay, all right, at least from the fluent side, where we seem fairly complete. So everybody welcome for another edition of uh the bi-weekly uh flint's community call.

A

Today, the uh the agenda is that we have an update from the core team which actually just came back from a week-long summit and then we're gonna have a demo from uh about self-id ceramic influence from fan from who won the bounty.

A

Hackathon question answers open discussions and then we're going to get a super cool lecture from pavel about the java sdk fluent for js developers. I think I'll hand it over to dmitry. Now.

F

Hello, so um just just a few quick updates, I encountered four uh updates to share today. uh First, I'm waiting for fan to to join. I think it makes sense for friend to join.

B

He's resetting his uh zoom because he didn't hear us.

F

Okay, uh so, first of all, uh we released the new version of aqua with the new version of aqua, vm and rolled it out to the network and to the fluence jazz uh implementation of the clean spear, and it's like it contains small improvements from like from from the usage point of view, but it fixes some really obscure bugs and issues that many hackers were facing uh it sculpted streams.

F

So previously, when you use the streams inside for loops in your aqua, you could have like met some really strange behavior of aqua, that it kept the state of the stream between uh iterations and now these streams are isolated and it means that you can safely use uh streams to build data within the functions which then will be called inside the for loop and that's just just nice and more intuitive.

F

uh That's a breaking change! So if you uh want to use it, if you want to have a more stable aqua, please update to aqua, 1.5 and uh fl. Distron is not supported, uh at least for now. So you have aquaran instead, which compiles and runs aqua in a single like cli command and be sure that you updated fluence.s and uh the node docker image as well.

F

So uh the second update is that uh after we rolled out the the asynchronous virtual machine inside fluence peers, um we like we are looking for the best ways to use it, uh and uh uh now we have a very widely used pattern uh enabled in aqua and influence uh its race conditions, for example uh the race with the timer, which leads to a timeout pattern.

F

So now you can uh isolate a block of code in aqua, uh wait for some time uh for it to be executed, and if it uh doesn't, uh then you'll have an error and can rub it with some uh failure. Logic.

F

So that's also pretty nice and we had a lot of requests uh about it and uh two more updates which are not uh in the dev channel, not not on the list of updates, but will be hopefully soon. uh The first is that uh there is a very important piece of this. Fluence security model called trustgraph, and now we have the truss graph live.

F

The service is already bundled within the fluence node, and you can you can try it out. You can utilize it. You can build your logic on top of it and that's pretty good uh and pretty important. It will have a lot of consequences.

F

um I think that we will have a dedicated trust graph uh related meeting uh in the following weeks uh to highlight the usages, but just to give a hint uh trust graph is dedicated, is designed to solve two big problems. uh The first one is a priorities for resources for shared resources in the open network.

F

So you can prioritize uh some peers or some groups of peers, uh or some people here are his to be sure that they will have the resources that you want to allocate to them using trust graph, and the second problem is uh labeling uh or um attributing uh appears uh with the cryptographic signatures cryptographic correlations, uh so this could be used for uh permission management, uh which is shared across different parts and protocols and applications uh on the flex network.

F

So uh we have it live. We have some examples in the code and working on the proper announcement and the documentation to provide in the dev channel, and the last is that we spent the whole week off on the fluence engineering offside event, uh which was pretty intense, uh and uh um hopefully it will be materialized uh in form of uh roadmaps for uh the fluence components and the much more fresh, wonderful content from the core fluence team.

F

So this is also just an announcement for uh what what's going on and what we will see soon.

F

I'm done with the updates there.

A

It was anybody have any questions uh as a bonus for those on the call I haven't updated our documentation from. I feel this to aqua, yet so bear with me for there too, if you run into problems, ping me on discord will show you how to use aquaron instead of airfield is okay.

A

Next up is fan. He won the ceramic hackathon around self id and he's gonna tell us all about his project. Fun, europe.

A

Yes, I can hear you can.

G

B

G

Okay, so let me show my documentation first.

G

Can you see my screen.

F

A

G

Sorry, sorry, my setup is not okay.

G

Can you see my screen? Yes, yes, looking good, okay, this is a documentation.

G

The user first start the authentication in the web, using the ceramics of eddy and after the successful, successful authentication, the id will be returned and using the sdid to request the greeting service through the all service, the all services in in the middle, it maintain a hash map.

G

If the the idea is containing a hash map, the user has a permission to access a reading service. If the id is not contained in the hashmap, the request will be rejected.

G

So the logic is very simple. Let me show you the example.

G

This is a main page and if we click the grading button because we are not authenticated, it is rejected.

F

Probably you you want to show something else. We still see the readme page, uh the window with readme.

A

Yeah, it doesn't show your browser.

G

G

The user first started authentication the web using the ceramic cell id and after a successful authentication, the id is returned and we request the grading service using sdid.

G

The all source is in the middle if the dd is contained in a hash map. The the the authors maintain a hash map. If the deity is contained in a hashmap, the permission can be granted to access a greeting service.

G

If the daddy is not containing the hashmap, the request will be rejected.

G

So it is, it is a very simple method to authentic the service. The greeting service.

G

So anybody has any question.

A

Yeah, didn't you have a demo on how to run it like in the browser.

G

G

See this is a mintage because we are not authenticated if we click the greeting button. We.

C

Cannot see the browser, we cannot see a browser window like you need to share the browser window of the screen.

C

C

Maybe just stop sharing and share again just pick another window and it should work.

G

Hope you, let me try.

G

G

Can you see the browser yes.

G

Okay: first, if we don't authenticate click the greeting button, because we don't have unsending, it will be rejected. It is showed you are not authenticated.

G

And then go to the permission, tab.

G

Okay, the click the self-id assembly.

G

And now we are authenticated.

G

Go to the service tablet.

G

Click the greeting.

G

It shows a high message with the id.

G

If we go to the premise tab, remove the pid.

G

And return to the service.

G

You should not understand.

G

The key is to add the did to the permission map in the all service.

G

A

Yeah, so so this is pretty cool. So why is this a big deal? I mean: what's the big deal about the self-ids and dids, particularly self-id from ceramic which you used what's what's about? What's the benefit of using that for authentication, and why is it important in decentralized systems.

G

The the cell id is a means. You can have your own data. We don't need to store the data with a centralized storage service. We started on the decentralized network.

G

I think that's a benefit of the using the dad. The id is a decentralized identity. No, when nobody can control it.

A

Right and- and but I mean one of the advantages of the idea is also that it's sort of one one key set for for different, basically universal right, because I mean self id itself integrates with a whole bunch of different uh blockchain wallets as well right. So you can use the wallet yes to to okay.

A

So that, basically, the entire key management now can be reduced to a wallet holding it, and then you can control in a decentralized manner any service or transaction submission to any on or off gene blockchain. Basically right right, okay, so that's that's! That's pretty powerful! Isn't it sure?

A

That's why I think what it is really cool, because uh it it really pushes that that authentication authorization level uh into into the vid space. Okay, anybody else have any questions for fun.

F

I'm just curious about the lessons that you have learned while working with fluence and uh if there is any piece of code that you have written in aqua or with marine, and if you can like share it and say something about it. It would be super interesting.

G

Yep, the the agua is quite a simple. The syntax is very simple and very easy to understand. I can, I think I have a programming background, can understand very quickly and the the theory, the the marine series.

G

I think it's also a different kind of docker docker service. I think because I use the docker before it is like a service. After I tried the strand uh florence, the marine service is quite quite similar with the docker service.

G

I get a feeling, it's a decentralized docker service, kubernetes, decentralized, kubernetes,.

G

And I think it's very powerful in the.

F

Oh future, you yeah.

A

A

All right cool, well, thanks fun! uh Well done! Thank you very much. This is awesome thanks for the presentation taking the time, and I think next on the agenda we have. uh We have uh question answers any questions, any possibly any answers. Anybody have questions about anything anything related, fluent aqua.

A

F

uh Fun you're still sharing the screen. Yeah you're still sharing.

G

Okay, let's be closed.

A

All right, silent crowd today, silent crowd! Well, in that case I'll hand it over to pavel who's, going to tell us everything you ever wanted to know and needed to know about java, the sd javascript sdk from fluent pavel all yours.

A

We can't hear you you're unmute, oh.

D

I'm so nervous. This is my first talk on the community called. I forgot them yet myself, hi bernard! Thank you. uh So let me share the screen.

D

ah Is it sharing.

D

Yep yep great uh and by the way I think the term uh javascript sdk is a little bit misleading, so we've renamed it everywhere to influence. Yes- and I hope this is my opportunity to tell me exactly why we did this and what should actually constitute an sdk for a front-end developer, who wants to develop an application for fluence and first, let's start with a little bit of description of the fluence, maybe uh for people who are just learning influence for the first time. Maybe they are viewing this video on the youtube.

D

So, basically, fluence is a peer-to-peer application platform uh and fluence provides a development stack uh with a lot of tools to help developers, uh building peer-to-peer applications and question them onto the network and by the way the network is live. You can develop and deploy services there and deploy applications for more infor visit. I guess our previous calls about fluency, uh mainly the introductory one. uh I will focus more on the javascript side of our stack, uh so fluences network consists of peers, obviously, and every peer runs, an aqua virtual machine.

D

This is basically a driving force for the network.

D

Aqua virtual machines are processes, particles and particles, uh programmable message, packages that are used to share data between peers and communicate between peers. So I hope my cursor is visible. This is like a tiny schematic representation of the network.

D

uh Every uh this square uh is appear and it runs uh by assembly aquavium and by the way uh years uh are divided into two classes. uh The rust fluence nodes, which are kinda backhand nodes running in the network, and we also have implementation for uh clients like browsers.

D

uh Maybe a mobile device is running javascript and stuff, like that, uh we call them clients and uh actually yeah this. uh This is what we are interested here: clients uh connect to the network, vr relays we'll talk about that later, and they are implemented in typescript uh just a little bit about their communication. So.

D

Our message is called particle: it cons it contains both code and data and the code actually encodes uh how the particle should travel through the network and what should it call from other from different gears um particles? Well, we can say, for the sake of this demo, that particles can be programmed with echo programming, languages, programming, language and uh here on this uh scheme. uh These green dots are particles traveling from peer to peer.

D

Sending some important messages- let's say like that, so about typescript peer implementation. uh It has everything uh needed to participate in fluence protocol. So, as I've already said, it runs a commercial machine. It has the pure id to identify itself in the network. uh It has the connectivity with the gs implementation of peer-to-peer.

D

It also has all set of built-ins required by the fluence protocol, so it can be viewed as a first-class citizen in the fluent network.

D

uh Unlike the rust implementation, which is like our uh backend or distributed cloud main computing power of javascript or typescript client has some differences. First of all, as I've already said, due to the limitations of browsers, it has to connect to the network via relay relay could be any node in the network and all the particles are sent to the relay and then only after that they are routed into the network.

D

Also, um typescript implementation doesn't allow to host marine services, basically services that are composed of wasa modules, yet because we're actually working on that.

D

Also, contrary to the ras node, you can directly implement services uh in plain typescript or javascript, and this is very handy for unfronted applications where you can provide some code for the callbacks or uh services that I run in browser very quickly without need to write a lot of rust code and compile it to us and stuff, like that, it's very user friendly.

D

Also uh typescript implementation allows you to actually send initiate your particular processing, so you can send the particle, and that provides a great opportunity for our user interfaces to provide the uh entry point to the network and the ability to to communicate with the network.

D

Also, we will talk. I will talk about aqua.

D

Akku provides first class compilation target for the typescript peer, and this with the combination of the ability to initiate particle, allows you to actually program a user facing applications with aqua and the typescript implementation of the pier.

D

Before we proceed, I have to show you a little bit of aqua and how it is translated to typescript for a more in-depth, intro introduction and more in depth info about aqua. I guess you should visit our previous call previous community call, and you can always refer to the documentation as well. uh Here we will uh be focused mostly on the typescript part, so, first of all services, uh you can define the service in aqua like this and for the aqua definition. The compiler provides you with two entities.

D

First of all, it's the description of the interface.

D

Basically, it is an object with the keys, mapped to the corresponding functions in the service, and it also provides a function which is called register and the name of service.

D

If you call this function, you will be able to register the service, provide and actually provide the implementation for the service, and when someone calls the service from aqua, your implementation will be get called.

D

uh Second is uh the function definition in aqua?

D

Oh, unfortunately, my cmx highlighting doesn't work here, uh but anyway, um here are two examples of functions in aqua, uh one which has a return value and an argument and another uh has nothing and uh for the functions that aqua compiler generates. You basically a function in the in typescript.

D

Everything is type checked and you have support from intellisense. For example, if you're using these code for functions which return values, they correspond to promises with return, value actually and non-return value function, state not returning the value and here's little uh there is a little catch. uh Please note that when you make a function, do not return anything uh the function, the corresponding typescript function will succeed.

D

Just after the particle is sent to the network, we will get to that later about particles, so it won't catch any runtime errors during the execution of the echo or the aquascript and the echo code.

D

If you really want to handle errors, you have to provide the return value of just any kind of acknowledges. Okay, and if the function has the return value, then it will catch all the errors that happen during the execution and the promise will fail with a user-friendly error message. So you have to keep in mind it when you are designing your functions, a little bit about types of the conversion between aqua types and typescript types are pretty much straightforward.

D

The only interesting part here are arrow types, so for every arrow type, the aquacompiler will create a callback in the form of a function.

D

So, for example, here we have an arrow of a string and uh integer and it will create us. The diaco will create us, a typescript definition of a function with actually arguments of exactly type string and a number.

D

So fluence provides, as I've already said. In the beginning, we provide the fluence gs package which.

D

Which provides some useful stuff for uh front-end development and, first of all, it is actually the implementation of the fluent sphere itself.

D

The fluent spear class has a public interface which uh you are free to call, and we also have internal interface. I will go to it later. uh Basically, it is needed to interrupt with the aquacompiler definitions.

D

Also, uh we provide a default beer. Basically, every time you call the function like uh this uh you're, creating the particle and then executing it from the default gear uh defined uh in the fluence.js package.

D

This is, I think this is enough for the vast majority of use cases and you don't do not have to worry about the exact year where the uh where the execution is happening, but.

D

You can always uh get the spear and uh get an insta for an instance of the fluent spear, and you can also create the pier manually, and you will only need it for advanced cases where you actually want to run many peers inside the browser- and I won't be covering this during this talk- you can feel free to refer to the documentation.

D

We have this section describing usage of several peers.

D

So, just quickly to finish with the athletes package server, you will provide the facade which wraps around this default peer and provides a set of common operations with the peer, and there is also a compiler support, which is not used publicly, but it is very useful and is needed for the aqua compiler to do its magic.

D

So this is the interface for fluence facade.

D

It is very simple. Basically, it allows you to start peer to stop it and to get its status. We are working or on that status and it will be extended very soon.

D

And speaking of the fluent spear public api, uh it is very similar to the facade one. Basically, it almost corresponds one to one to the first aid api, but uh actually the facade api uh calls these functions from the default implementation of beer under the hood. There's also one.

D

One need function which can help you to check if any javascript object is an instance of fluent sphere and it is actually used and the aqua compiler support as well.

D

So uh let me tell you a little bit more about the affluent sphere internally, so under the hood, the fluent spear, as I've already said, it has an aqua virtual machine.

D

uh It also uses lip22p for the connectivity with the network and it provides something like a service registry where you can, where actually all the service registrations are stored and from where they get called so everything is connected using.

D

Using queues or I'd rather say, rxgs streams, if you want me to tell more about it, feel free to ask any questions and regarding the internal api, the one which is used by the aqua compiler under the hood, so, first of all the initial particle, this is the magic oh and by the way yeah. I guess it will be.

D

My demo will highlight it for me, so initiate particle is where the particle initiation starts and what makes it possible to call all the functions made by the aqua compiler, and there is a a service registration object. Let's call it like that.

D

D

Under the hood, we divide services like all all the common ones and the ones specific for particles, but they are more or less the same, just some magic to help the compiler pick the correct service. When the particle comes to the beer, there is also a handle for callbacks and with all of that, the compiler can generate its functions.

D

So, let's uh see how it works under hood, for example, uh for a function named, say hello. The compiler generates, first of all, it generates a set of overloads, corresponding to different type of arguments passed to the function.

D

As you can see, you can pass the peer as the first argument, and that way you can execute the function on a different peer.

D

There goes the actual implementation and, under the hood.

D

This call function is called, and this is where the glue code resides the glucose between aqua compiler. Well, this this definition here and the actual conversion to particle and particle format. This is where it's done similar to that um for service definitions. There are two: there are several entities that are generated. First of all, it's the interface, it's very handy to import it in the application code, and then there are several overloads uh the same way.

D

You can always pass fluent sphere as the first argument, and then there is the actual implementation which, under the hood, calls the register service function and that's roughly the same as the call function, but for service registrations.

D

So now, let's talk a little bit more about how you can actually build something with fluence.js and that's. Why that's where I should talk about the sdk, so basically uh building blocks for a front-end application uh mania. First of all, let's uh the aqua code, you will need to write peer-to-peer communication logic.

D

Also, you need the compiler cli. Well, you needed to compile a code into a typescript. uh Of course you need the type the fluence package with the which provides everything I've been talking about before.

D

Of course, you will need the javascript environment.

D

We have both examples for node.js and browser. Our is the case tested and it is working in all modern browsers.

D

So you are free to choose the environment for you and uh of course you will need to write your application code and there are no restrictions for you there. You can use any framework, you can use it without framework, so I would say that the sdk for a javascript influence consists of at least uh two things: that's uh the aqua compiler and that's uh the fluence gs package itself, and I'm not sure, but we can probably uh consider the aqua code uh somewhere close to this uh to this sdk as well.

D

So the whole sdk is a set of uh tools, but uh fluence.js is actually the package which uh provides you uh the implementation for the part of this of this tool set.

D

So in order to write application, you have to install the dependencies.

D

One interesting dependency I want to talk about is following network environment. This is a package uh where you can find lists of the nodes of a known fluence network, for example our krasnodar network, which is kind of like our most user and front-facing network.

D

All the examples are pointing to that, so you can find it in the fluence network environment package. Of course you will need the compiler and you can install it as a development uh dependency. uh You don't need uh the code, this code in production, but it is needed to compile aqua at the compile time and we highly recommend automating the compilation and uh we use chocolate dar in our examples. So uh we will have a chucky dar in this demo as well.

D

So after you install the compiler, it can be called with an npx, but it is much more convenient to have separate scripts like this one for to just call the aquacompiler, and uh we use shortcut to watch uh the directory for the changes in aqua files and to run the completion automatically and after you define such to define such scripts, you can call them via npm, okay, let's.

D

Let's have an example, so this is knuckle code for our application. uh Here on the first line we are importing built-in functions, we will need them.

D

uh This is the service definition it will here. The service will be run in the browser and it has a simple function which has an argument of appear where, where the call came from- and it returns a stream- and here is the function- which does uh a very, very simple thing- it uh goes to the target pier uh through the target peers relay and calls the service we defined above and returns. The result back.

D

So if you compile this code, you will get the typescript file and, let's see how you can use it in the application so first of all uh well well, I want to say that the this is the demo uh with the react, uh but it can be done without react, as I've already said, with other front-end framework or even without it, so uh I will focus on the fluence part. So, first of all, we import fluence, uh the fs8 uh I've been talking earlier talking about earlier.

D

uh This is the nadar network. I've been talking about. We import that from the known packages, and uh here is where we import in functions generated by diaco, compiler.

D

So this is how we start the fluence node. I mean uh here's, how we start the default gear, the sphere which is right inside our current browser, and we are registering a handler for for the hello for the hello of the hello world service. So we are just setting some hello message somewhere in the input and returning back a welcome message.

D

And this is actually oh I, this is the incorrect line. I wanted to highlight this one. uh This is how you call the function. You call aquacompile functions just like any normal, asynchronous typescript function and we are constantly uh say hello, uh providing it with some peer id and it uh and its relay.

D

uh We get them from the inputs, and I will show it to you in the demo so demo time, uh I've already opened the project, and here is our code. Everything is already installed. I just wanted to save your time. uh The only thing I would like to do is to compile aqua.

D

To show that it is compiled uh here is uh the generated output from the aqua compiler.

D

This is the interface for our service, overrides for the radio for the service, registration and here's the implementation for the say, hello function. Actually, this script, it is called air which stands for aqua, intermediate representation, and it is the part of the particle, and this is the code I've been talking about in the beginning.

D

This is actually our application.

D

Let's, let's just start it.

D

Okay, so we have two browsers: each running our application: let's connect to different relays and we want to say hello from discipline from this peer to this one, so we copy the corresponding hero, ids and relays, and here it works.

D

So getting back to the demo actually, uh this was, uh I guess, the introduction to the fluence gs uh feel free to refer to our documentation.

D

uh There is no link to this presentation, but I will, I think I will upload it somewhere and we will attach the link under the youtube description, because I think this recording will be only on the youtube, so you'll be able to play with this presentation as well.

D

So, thank you for your attention. Any questions.

A

All right thanks, pablo questions. Anybody.

A

All right, I got a question for you just in case I saved this up for the last couple days uh and uh and the rust nodes have a lot of built-ins functionality which we can easily access from aqua, and I was wondering if those capabilities are available from the js based nodes as well.

D

Well, there are, I guess, several type of built-ins, some of which correspond to the marine. That's about creating services, uh uploading modules and stuff, like that. Obviously, gs implementation doesn't support that, but all the built-ins like identity or timeout and uh stuff like that they are fully supported. So you can use them.

A

Very cool so most of the peer stuff and some of those are all available right. Yes, of course, okay, cool yeah, just double check it. I think that's that's an important point, I think of uh of when, when when, when we look at it, that uh there's a there's, an equality or an indifference for aqua, whether what node you're hitting on right.

D

uh Yes, of course, uh it is different uh and you can see this difference.

A

No, but what I mean it's from from for aquat does make a difference. So if I'm, for example, if I wanted to uh use pure timeout or use.

D

Yeah, I don't need to know what.

A

What what type of node I have in my neighborhood? I just use yes and it: okay, cool, very cool.

A

Anybody else have any questions.

A

Well then, thank you dmitry. Thank you fun. Thank you, pablo it's awesome and uh for those of you celebrating thanksgiving, happy thanksgiving for the rest of you, too bad.

A

And we'll see you in two weeks right anna.

A

All right cool thanks: everybody have a great week, have a great weekend, see you in a couple weeks, bye.