Fluence Labs Community, 28 Oct 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Developer Community Call #4 – Marine

Description

Today we will talk about Marine — a multi-module WASM runtime which runs AquaVM and services on Fluence nodes. The main feature of Marine is a shared-nothing linking scheme: modules can export and import functions from each other but do not have access to others' memory. This was achieved using interface-types. It also provides controlled access to the filesystem and system binaries. The main focus of today's talk is building a marine service using tools and SDKs.

00:00 Intro to the call
01:10 Context of the call
05:55 Describing Marine service (service design)
10:29 Implementing service (curl_adapter module implementation)
13:47 Building the curl_adapter with marine tool
15:12 Config.toml describes how to make a service from modules (service configuration)
17:58 Looking at the built module with the mrepl tool (testing with the mrepl)
19:36 How to access the network from a wasm module? (using mounted binaries to call curl)
27:39 Full marine runtime in mrepl, Q&A
32:57 Writing module for filesystem access (using WASI to access filesystem)
40:09 Writing of facade module which composes storage and curl_adapter. (importing functions from other modules)
47:40 How to use marine_test
57:47 Why marine_test is useful
58:56 How does other project use marine_test
01:01:30 Updates from the tech team
01:03:42 Q&A
01:06:31 Morgan Moskalyk demo
01:17:17 Feedback on Morgan’s demo
01:20:15 Q&A on Morgan's demo, Morgan and Joera discussing using Fluence

Speaker: Valery Antopol

Useful links:

Fluence documentation https://doc.fluence.dev/docs/
Discord Support https://fluence.chat/
Telegram https://t.me/fluencedev
GitHub https://github.com/fluencelabs
Join Fluence Labs https://fluence.network/join.html
Community Calls wiki https://www.notion.so/fluencenetwork/Fluence-Developer-Community-Calls-2733beddb3ff4e4986c4fcc830fc796d
Fluence Developer Updates https://t.me/fluencedev

A

I don't see bernard or bernard is here.

B

We usually have an intro from you and then some same related topic. So do you have any intro about.

A

Playing something.

B

A

We have some connection issues I can cut off at any moment so.

A

uh I don't know, like I mean while there you can start uh if you want or if anyone wants to. If anyone has anything urgent, don't hesitate to share, and I guess the plan is as as at previous calls. We're. Gonna have some for some talk about uh one of fluence components uh by the fluence collapse team and then q a and whatever everyone would have has in mind.

A

B

All right, so let me start today I will talk about uh marine. It is uh our web assembly runtime, which we use to run services in the friends network. It runs the services it and it turns the aqua vm it which is which runs the which impress the aquacult. So they it will be not the talk about internal implementation or something due to the like a demo like. I will create a service using the all the tooling we have, and so do you describe the experience and the tools I will use.

B

So, let's describe what the let's then just start going to the ide.

B

And share my screen.

B

Great, do you see.

C

B

A

So you maybe can zoom in a little bit. But yes.

D

Make the fonts bigger.

B

Microphones, bigger, oh okay: um let's talk no.

B

E

There's presenter mode is the simplest way. I don't see that in the view or triple shift you can.

F

E

Triple shift and then presenter mode.

B

Triple shift press and there is in.

E

B

E

Oh and something else after that.

B

Something else.

E

So it wasn't the best advice, probably.

B

You should run enter presentation mode, no.

B

um That's not the thing and.

B

It's not that what I wanted.

E

Exit presentation mode yeah, there was a presentation.

B

So, let's just make the phones bigger- maybe maybe that's fine.

G

Maybe just the phones are completely fine for me.

B

Let's make it like 18 a bit better, maybe but I'm looking at like 13 and.

E

Press apply close the window and go ahead.

B

B

A

E

But probably it should be in appearance.

B

Let's start without it, I think it'd be just to spend time or nothing and do anyone want the phone to be bigger.

E

E

For me, it's nice, I think evgeny can survive. It.

A

Yeah, I will survive yeah. Sorry.

B

Yeah, okay, so imagine we want to have a service which will download something from the network, then save it on the disk on the file system. So it is a simple example which already can be split into the into some modules into some parts.

B

So we want the service interface to be some function. I will write it in rust, like.

B

Let's see at the interface, I want to implement.

B

It can be just rail rail in the form of string.

B

And some file name in the same form to.

B

B

So let it be the service interface and we can split the service in several parts. We can create a part that goes to the network, create a part that goes to the file system and the part which composes it imposes them and thus of the internet, provides interface of the service.

B

B

Marine is there.

B

So, let's see how how can we divide it? We can create a service, the other model that goes to the network, let's call it serial c url adapter.

B

Let's have a model that goes to the file system, let's call it storage.

B

And that module that composes it and provides interface, that's called facade as it the model that looks into the outer world.

B

So what interface can we provide in the provide from such models like serial adapter, can be and provide a function download which will take a your url.

B

And return something we don't think here about error hanging and anything such so. Let's assume that anything always returns. The results.

B

It's home and storage can go to the file system and save the file, so it will have the function like store.

B

Which takes file name and some content.

B

Okay, it's and it doesn't have to return anything and let's have the functioning facet that we already described. Let's go to state so and marine allows us to implement. Everything is different models that modules then link it and run as the one. So, let's start implementing, I will start with a c url adapter.

B

Okay, let's create a great.

B

Okay, let's add it to the workspace.

B

So we have the same target directory for everything.

B

Okay and we need to add the uh marine marine rust sdk as a dependency, so we can write our model.

B

Okay and we can start writing, we usually need the main function as uh runtime calls it when it needs to initialize some of the functionality.

B

So, let's keep it empty.

B

There we need to write a function that, as we described, that's the download thing.

B

So, let's see how we implement it, we need to wrap it with the marine macro. uh In order to tell uh the comp tell our runtime, which functions we export to the interface. So let's wrap it in the my marine.

B

Okay and let's just try to return something, don't go to network just.

B

Make something.

B

B

B

If we can compile it, we use the macro from the marine.

B

B

So here we have the dependency, we have to find macro and we have a function. Let's build this crate.

B

We have the marine uh build helper, which.

B

Distributes in cargo, which can be downloaded by cargo like this.

B

I already have it so I don't need it to install. So let's try to build our module.

B

That's button release, you don't need the debug, so it will download all the dependencies from the marine rust sdk and at an end, will build the.

B

B

So we build it: um okay,.

B

um Then we may want to try to call the function and the let's and loop doesn't work. Does it work to create a service? Actually, we need a config that tells which models modules are in the service, so we create a file.

B

So we create a file.

B

Let's call it config.tamul as it in there must be in the thermal.

B

Format, so we need to tell where are the our modules.

B

And they are in the target directory.

B

So I think, hey, I didn't add the project to the workspace I forgot to do it so.

B

Let's build it again.

B

It will create a person as it usc uh directory target directory, so there will be our passing file and we can provide this path.

B

To the config file target then awesome: let's see.

B

Please, and so.

B

Then we need to provide some need to tell the marine which models are there are in the service. So so we need to create a module part and name the module it till the name of the webassembly file like like.

B

Like this zero adapter awesome.

B

So, okay, then, uh to test our service, we can use the marine ripple in short and repel the cli tool to inspect the service and modules, so we it can be installed as marine in by cargo.

B

So, let's look at this service.

B

Provides here the config file.

B

So this is loaded. We have some instructions here. They.

H

B

Pretty simple: now we need the interface instruction, a look at our module.

B

So it loaded one module serial adapter with the function, download text, strings sync and return string, and you can call it as their mineral has the marine run time inside it? You can use, call and provide the model name.

B

Then function then argument. We don't need it actually.

B

So we see it works, we got something. Okay, then we have to somehow go to the network. um As you know, webassembly doesn't provide direct access to system interface and such things, so we need to provide some way. We needed to provide some way to access serial, for example, a common line tool.

B

So we implemented a the approach we call mounted binaries we can in the config file, provides a.

B

B

Called model dot mounted binaries.

B

Then expose some names here like zero and write a path.

B

It is in user, I mean serial for me so and in the source code we can import the uh there. This command line two as external function, so we need to write a stern c, lock, wrap it in the marine.

B

And also tell from there we need to import functions from this external model and for the for the mounted binaries. It will be the cost so link and provide you with version import.

B

Then, if we provide here the cost, then we will import binaries and not functions from other modules. So here we will have a function named serial, as in config we should take a vector of strings. The online arguments.

B

And which will return a structure from sdk called mounted binary result.

B

And that's it: we can uh the marine uh wraps the call to the.

B

Binary on from file system uh into the function and provides this function, other module, so we can use it in our services.

B

Granted binary result consists consists of some useful fields like return code of the program, uh return code of the in the sync form there out and stdr, and so so, let's return to our downloader so and use the serial here. So now we need to use url.

B

B

The idea highlights the call to the serial as a because things it is unsafe, but actually it will be wrapped by marine, but when it will build project, so it will be nothing, no warnings, no errors about it.

B

So you have the result. We don't care about errors, anything. We think that the result is in std out as usual. So let's create a string from the results.

B

Let's think ctf eight.

B

B

And we can unwrap it because we don't care about the errors. So now we can think that is ready. Let's go to the directory.

B

And build it oops reverse.

B

It is strange because we just built it without any problems, and we.

B

F

B

Okay, let's see if we have here, we heard here.

B

um Onto the binary result, I heard here and you can just undo it okay, so, let's again and it works.

B

B

Okay, it was stranger, but it doesn't really practice anymore. Well, that's just it.

B

And then ripple have the same function and ripple stores the last the star's instruction start starts input history, so you can just run the same string and it will return nothing because test is not a valid url or something so it will be an error. Let's go to the, for example, google.com.

B

And we will go some return. We didn't provide zero with the additional flux, so it returned first result and it was the description that we have to go somewhere.

B

Okay, but we see that the idea works. We actually we really did go to the network, and so we got some result. Okay,.

D

Let's valeria before you go on, can you explain something here, the the rebel that is the actual marine runtime, that's usually a node. It's basically running on your local. Is that kinking like explain a little bit about that.

B

uh About the rain on time in the in the m yeah.

D

The marine run time sort of a little bit how it does. You know how it takes the code, how it takes those modules, how it processes it and how the replica relates to it locally to the remote or.

B

The deployed one uh we will not deploy that service here to the network. We don't need it, it doesn't just uh the tools so about the.

B

uh It's just that we have a runtime. The lab is a library and we use it everywhere when we need- and uh we have one difference uh in the interfaces of the I'm rebel and the um and the node. That nodes has the such uh the concept of the state module that uh that, uh if the service consists of several models modules, then a node shows under the interface of the last module in the config.

B

So uh for all, underlying modules are hidden and apple provides all the interface to all modules, so it internally it takes some modules and.

B

Runs each in the run time in each module in the separating box and.

B

Which has only interface that is given to this module in the config and we'll talk talk about it later.

B

If the module has, if we have a multi-modal service which can link uh which can use where models can use functions from each other, they're really be there.

B

The enrable main and time in there will link the functions and structures of modules internally and to provide a safe way for the data between modules. So the marine on time provides that shared, nothing linking scam so and one module can access memory of the different module in in any way unless the the host uh explicitly.

B

uh Given this permission uh in the some through some workaround like provides like memory mapping from somewhere like it is hard to do accidentally.

B

I don't know what to you. What do you want from me more about what I can tell you right now.

D

So one thing you did was over the last week you created a testing, a multi-module testing unit right yeah and that basically allows people to test their code.

D

There was modules within the russ code as opposed to having to do it. Many in the rebel right, that's sort of one of the updates.

B

D

B

The way to that to test uh without an apple but automatically run uh loads was some models into the environment into their runtime and uh use and access these functions from the last fold uh for the single module from the for, for it works for the single service, we can use different service. I wanted to talk about this laser, but I think we are short in time so I'll try to be slightly faster this time.

B

Okay, so we have the model which goes to the network and we create a model that goes to the uh file system. So, let's create it.

B

Let's call it storage, okay, you can add it to them workplace and then.

B

Add the same dependency here, so we not need this and we need to add another module into the into our config service. Config.

B

And it will need to go to the file system, so we need to provide some interface to say where it can go, because if we don't say it will be, it will not be allowed to go anywhere in the file system. So we have used the webassembly system interface for it and.

B

Provide some we can see which directories it can reach. Let's say.

B

We have a updated year which will map into a better deer internally.

B

And we can reopen it.

B

B

So, okay, it will this line describes, which directories uh we are allowed to read from the boston.

B

The proceed ensures that we don't treat anything outside them and this line says that we have says to the time that we want to file form to open files in this directory and this directory before running, not lazily, but at the start.

B

So let's use it. We can just use a common file system interface from the rust to use the file system. So.

B

B

First, so off.

B

B

B

Let's implement the function, so we need to write some boilerplate.

B

B

To make a full path.

B

This will be data and file name.

B

Then you have to write to create a path actually.

B

I forgot you write choice, okay, it might not compile, because I don't remember the view or from I should use for the path then you'll be able to store and we will just store yeah they're, doing good.

B

B

To see if there to have a panic, if there was an error and just dropped result: okay,.

B

Let's build it.

B

Oh, I forgot to wrap it in marine.

B

So, okay, we built there.

B

um The model and we can create directory, we can test it with repo.

H

B

Maybe we can just.

B

um They create a directory in the right place, no.

B

B

I did, I didn't create. Actually it's, okay, everything is opened. We have two models right now. I can call storage model.

B

Storage or and then we can provide several arguments in a ripple. The arguments are in json, so it is an array of two entries, two things or objects or something um we can pass here like some file and then some some content.

B

Let's refresh it and we have some file with some content, so it works great. Then we need to wrap both of wrap it into the one module which will provide both. We need to compose it. Yes, this part, okay,.

B

B

B

B

B

Here we need to link to import functions from our modules.

B

Can do it using the same pattern with extern.

B

But this time we will provide the name of the model from which we import functions.

B

Okay, so then we can just copy the function. Interface from there.

B

Then this and this okay, then we want to impart the store function from the the storage module. Then we can go to the storage and store and a name and content.

B

Okay, so now we have include imported our functions and we can write the functions, final function and note and say downloaded store. Finally,.

B

It will take there.

B

And it will do there simple thing: will.

B

Rule download and save, as expected.

B

B

And wrapped in marine, so it will work.

B

So it will, it will be in the interface okay, let's go to the facade.

B

And try to build it.

B

Great we build it, go to the config at face 8 here it will be last. If we deploy it to the node, then on the last module it will green interface. So password must be last.

B

Okay, now we can use a maple.

B

See that another model up here and call the function.

B

And let's go to the google.com.

B

And store result in google response.

B

So I think I named it differently.

B

Yes, though, okay, let's just change the name, nicole.

B

B

And we can see the result here, so we composed here some modules into us into one application, which we call a service which could be later deployed turnover.

B

But now we are running out of time and.

B

I think we will not.

B

Have time enough to show the testing sdk I'm sad about it, but that's it. We looked at the marine which will which builds our uh models. We looked at the marine sdk, which helps us uh describe the interface of the model and looked at the m repo, which helps us test. Our model by hand.

B

So I think that's it.

B

I don't think I want to say anything more, but I think we will answer questions if you have any so, let's ask it in the voice, chat or in the text chat and to help you with your questions so I'll I'll wait for it.

D

So, yes, I have a question loaded questions so in this multi-module uh uh environment.

H

D

Of using the repo, we can now test or run those uh multi-module tests in in the code. Right I mean that's what you did over the last couple weeks: yeah yeah and.

B

Maybe I don't know if you could just quickly bring up the.

D

Marine example the multi-module example from people and just show us what it looks like on on how to test those in in code, as opposed to interactively.

B

So, okay, if you really wanted, I will go for it. That's pretty! We have we created a tested testing indicate which takes our past modules. It doesn't take given code of it doesn't need human code on the services and it takes the interface from the washings and provides it to the user, runs internally the marine on time with the buses and provides trust interface for it.

B

So, in the common rust tests we can write the code which will automatically run from time and all some functions and do everything we need, so we can do it in every crate, so I'll do it in the microwave.

B

Just uh one simple example:.

B

First, we need the switch to the nightly compiler, because test sdk needs nightly.

B

Great uh then, we need to add the sdk to them their dependencies.

B

So, okay, then we could can just write tests.

B

um Regarding the thing that you are not mentioned, but what I did in the uh last week, it is not the multimodal testing multimodal testing revised several months ago. It is multi-service testing, we can load several services.

B

So, let's see you can create some services just by copying the config tunnel and providing different modules. There.

B

So, let's do it.

B

B

So we can select everything we don't need here enter anything we don't need from here and we will have like two services.

B

So then we can use our running test. I can use it.

B

I don't have the.

B

Okay to work, so marine test is a macro which wraps a function of the module right now it will use it around the function.

B

um Let's name a module.

B

It takes several pairs of the modules here and config file and they are named so we can access.

B

We need some time to show it.

B

B

Okay, I have written the large tributes for this function.

B

Let's write up this function. The marinette's macro provides a marine test marine test kind of model, which holds all the interfaces.

B

Then we can create service.

B

It is a micro, so we don't have any completion here, but I remember anything: easy storage need here: storage.

B

So each line of the new of the service interface creates new, runs neuron time and creates new service. So then we can just do the same, that we did in fact model button tests and imagining that server serial adapter and storage at different services.

B

So we can call the service.

B

B

B

B

um Then we can just move here because we mutate the internal state.

B

Then we can store.

B

We need dust to string here because it because the interface extinct, not the reference of the slice, then we can treat, can read the test file and try to and check its content somehow. So, let's use first, let's do them first path. Okay, then,.

B

B

B

B

Okay, here we have me new okay, then we can read the content of this file.

B

And, for example, check if it is not empty.

B

Or check if it equals the result, which is better you just need to clone here then.

B

B

B

So then we can run the tests we can run is as used and as usual, with argo test.

B

And we release usually because it's.

B

It uh has a lot of fault independences and uh runs much faster in release mode than in.

D

While we're watching that compile and in the in the in the essence of time here, this is obviously.

E

D

B

D

B

Know I know I know, but.

D

In your words like, why do you I mean you spend all the time? Why is this really really helpful for developers to.

B

D

B

It is very helpful because the only way that you can test such thing in, if you have some several services, are not supposed to be linked in marine, you can only deploy them to the network and write some aqua code to compose them and.

D

Right so now you can do it locally. Instead of having to deploy everything. Now you can you only deploy once you've tested locally and you get the full test locally, as you had before. Only on the network right? Yes, that's a big freaking deal.

B

And it can be done automatically.

D

B

I

Cool and also every team who like get used to this issue, and it's uh when we're developing, there is only one ras developer in the team and when he produces something he can uh imitate some flaws uh in in this approach and then other developers uh who can take different decisions. What to do with these services in aqua can write aqua later, but they are sure that under layer in rust is well written.

I

So it allows a better split of worm between developers than if you write aqua, because aqua takes set of different decisions when you implement it. So it's very useful yeah.

D

All right see it's still compiling, so we use the timer.

B

D

Yeah right are there any other questions from anybody.

D

Otherwise, if guinea, you want to wrap it up,.

J

um No because we have also morgan today to present.

H

J

Maybe evgeny.

H

F

J

About the new format that we are we're talking about right now to everyone.

A

Yeah well, actually, uh we realized that uh just right away, starting with uh lectures and getting that influence, is kind of hard for everyone, so we and we sort of missing. Also the public updates like what's really happening like what uh fluence labs uh team uh have delivered uh since last quarter. In last two weeks, we were a little unprepared for for for today. I guess so we'll try to make it better next time.

A

Likewise, uh vice versa, I don't know if uh anyone from uh fluence labs wants to share any updates just like do it uh and um we also yeah. We also should have presentation today and yeah and next time so next time we will do vice versa, we'll start from updates q, a uh some demos and and then like whoever have time uh and desire to participate in a longer educational session and we'll be able to to do it so.

E

I can share a couple of updates just quickly, and probably I can oh no. I will not uh turn on my video. So on the last two weeks uh a big part of the flint's team was in lisbon.

E

uh That's not a technical update, but uh was in a is lisbon conference and around and that that was pretty cool uh and uh speaking about the deliveries. The main delivery over the last two weeks was a aqua 0.4 and it targets asynchronous javascript.

E

It's briefly described in our telegram channel, but like previously, all the aqua code on a single pier was executed. Sequentially, even if you were writing a parallel code in the script and now on the javascript side, it's executed in parallel and asynchronous, which is kind of nice, and we are working on doing the same for the node for the rust implementation, the fluence node and it's on the way. Hopefully we will have it next week, but not sure, and a lot of bug fixes all around just all around.

E

So that was the quick update.

G

I just wanted to add a little bit about the asynchronous javascript side. Basically, previously, you were only able to write synchronous code in the handlers and now you can use asynchronous code. So uh if you want to call some api which uses promises under the hood, you can use this api inside your callback.

E

So we are done and ready for the demos or if there are any questions regarding anything um influence, they could be asked right now or later or like.

F

I guess I had a can you hear me.

E

H

F

F

Cool, uh I just had an initial question regarding I know: there's some uh hackathon ongoing uh with git coin and some other things uh wondering if there's gonna be an idol like. I know that ceramic there was one with people using ceramic, possibly using ipfs or like textile, I'm wondering if it's going to be automatically out of the box implemented.

D

By automatically out of the box implemented, you mean like ipfs as a built-in service on the node, uh not necessarily.

F

Just some type of uh I guess some type of identity service, that's that you can call from the from the marine module, or does it always have to be external in which the person or developer is deciding on which store and keys are getting used.

D

So right now there are no plans for a generic uh did. So it's it's quote, unquote, vendor based right, so we have the from the last hackathon. We got ceramic self-id integrations and uh there were a couple submissions that he actually did it from marine as opposed to uh the front end, so that was pretty cool, but uh uh despite the id being a standard, it's really implementation, specific, so also on the resolver. So there is no plan right now from us for for a generic implementation.

D

But if that's something, that's uh that's a tooling aspect. That's important! We certainly are willing to entertain that that build out. So if that's something you want you're interested, not just contact me on. uh You know the emmy or anna or whoever.

A

Okay, thank you.

A

So morgan did you want to present uh anything to demo anything today or or you're good, and you mentioned.

F

uh Yeah I had uh some initial stuff. I'm everything is kind of a work in progress from the previous if online hackathon, uh but I've sort of started to strip things down and rebuild sort of a an alternative uh to the existing uh hack that I was working on. uh If you would like me to go for it, I will um basically I'm coming in more from a higher level perspective from the application.

F

uh Point of view and uh haven't necessarily focused too much on the marines in the aqua side, um japan example and I've slimmed it down a little bit uh and built an application surrounding the use of non-invasive brain feeder interfaces into a type of group.

F

I guess I might call it like a vibrational rhythm uh interface, but I have some initial questions and hypotheses that I'm kind of shooting out there to gain some perspective from from you guys as to where these technologies from a peer-to-peer perspective might help other domains. So I'll share my share my screen. Let me know if you can see.

H

It can you see my screen.

B

We see that screen has started screen sharing uh text hi. Can I still hear me.

A

Oh no, it's here, yeah.

F

Okay cool, so the project was called arbor, uh but there was a few mods or sub abstractions that I was looking at one. I was calling the fawn lore, and that was the um the wrapper to the um to the aqua uh or just the fluence network interface, um and then the other one was called fog facts which is like a basically taking the data that I'm sending across this network and compressing it into a type of visualization. That would be a possible output for nfts.

F

uh So I had certain questions around uh these technologies. One was what might peer-to-peer data connections feel like where data is being sent as particles in a type of virtual social network performing.

F

I read a paper once called mix hop where certain orientations of network neighbors were able to. I guess, collate the way that their identities are formed in a reputation fashion and by doing a type of mix-up or randomness. There was able to be better results on certain, let's say our intelligent applications.

F

So I'm wondering if there's a a solution as it relates to the spectral shapes for communicating with something like I'm wearing a brain, non-invasive brain computer interface called the blueberry, which I can go more into it later on, but basically it just takes. um It uses your blood flow as a proxy for brain activity.

F

So I'm wondering, if there's sort of ambient health state computer-like calmness that we can get to, if we're, uh let's say finding optimal uh neighbors that were vibing within the same frequency level and how this perceived randomness from let's say, cognition can be used for staking something like a biology like proof of attendance, on proposals for crosstalk communication.

F

So the premise is, if you have different dows across a network, let's say in you know, using normal blockchains can fluence network be the sort of in-between space of which these particles are being sent, and it's cool to hear that there's asynchronous programming in the javascript front end to do this, because I was butting heads with my web browser with all this.

F

These clocks and things running in my example, um and then my next question is: where will the active medical and the dreaming neuroeconomies be created with these types of nft compressions um paired with something like an ethereal cognitive study, uh whether we're talking about metaverse uh in sort of like virtual worlds or like our risks?

F

And my last question is: who will participate in possibly solving the problems of things like carbon life form conservation?

F

So arbor, as a background, is meant to solve the problem of reforestation with the use of a type of portal or interface. That is uh like an alternative reality game that one can use. I won't go into today, but I will just show the kind of like minimal flow uh or just the sort of segments that I see with these technologies working together, where we have our collective vibrations.

F

Where we're summing our our results of our cognition into some type of interface uh as an nft, we can possibly map that across some type of geolocation we'd be getting. You know, new shapes of formations from these data particles or wavelets that are being sent across the internet and can sort of map these uh against nature.

F

um There was a proposal done by vitalik talking about cross-roll-up, nft wrappers, and I think that there might be some type of solution or question to be solved as it relates to if these nfts that are being aggregated in these types of radial formations um are sort of like true to what we can sort of pass along to each other. Let's say from group to group: let's say squad to squad or, like trust, network, to trust network.

F

Can we also align around some of the other l2 solutions that are being implemented across the different projects to accommodate this and whether uh whether the fluence network that pdp implementation is being used, or you know something in the in other ibc networks, etc? um That's sort of how I kind of picture things. It's like. We have normal blockchain nfts, that's sort of your name, typical tree structure. Then you have the air interpreter as sort of the intermediate computing representation of these wavelets from some type of cognitive device.

F

So I'm gonna just have a minimal demo. For now things started to break because uh I was working with sound on my computer and trying to visualize things. But basically the premise is: if we had let's say four different people that are connected across the fluence network and their data is being sent from their local from their local device across the network and being aggregated at sort of local points.

F

um Can we get a sort of a radial fashion? Have that as a nft output? So again the um the example I I used was just a slimmed down fork. I basically removed the in-memory history of uh the messages being stored in the in the fluent pad, because I wanted the adapter of using textile as a data storage uh at sort of the local endpoints, not necessarily across the fluence network nodes. I think that'll have better capacity as it relates to how many messages can be stored, because I was hitting some some barriers with that.

F

When I was doing my example join the network and.

A

Press timed out for.

F

Join just give it a quick refresh, there's just one extra page.

F

And apologies: I'm uh in a bit of a not the greatest working environment but wanted to be here for the community call.

F

Okay, that's another shot.

F

So I go to the next screen and it's asking me to connect a bluetooth interface, so I'll scroll down to my blueberry and pair that right now it just has online s0 and it's loading.

F

It has my previous offline users being loaded in and we should just get a visual where, basically the premise is, the uh signal is being sent across the network. I'm performing an fft on these windows of information, which I'm then aggregating and you see in the bottom here.

F

Some data- that's that's, being accumulated from my cognition and then this part has broken for time being.

D

F

I would click, and I would have some type of audio signal- maybe mapped on something that actually sounds much better and not so annoying, but the premise is to sort of get like a genetic dial-up for what a cross-internet yeah genetic dialogue is kind of like the coin or the term I'm using for the time being, um but you get this radial fashion where the higher frequencies at the center, the lower frequencies, are at the outset, um and it would play a sort of tune and that would be sort of a nft output from this.

F

um So that's my presentation. I just want to say you know, thank you so much uh for all the incredible work that your team is uh doing. It's I find it really cool and so far I I hope that uh there's some additional questions that may be posed to the team, as it relates to some of the use cases in the sort of uh possibly medical domain, uh that I think that this technology might be applicable to so. Thank you very much. Stop sharing my screen.

A

Whoa, that's uh that's really amazing. I mean that's super great concept.

A

It's really worried. I like that. It's really worked like I, we heard the sound. It was like a little uh hard to hear because of noise cancellation uh that probably zoom enables but um yeah that's a super cool.

A

F

I I did have it working for it. It was like fluctuating based on the uh based on some notes.

A

Oh, I guess maybe it's my connection or not my connection based on what.

F

Probably my connection.

F

Well, I'm open to any questions. If you can still hear me, if not, I can just connect on telegram group really thankful to have that available to the developers.

A

Well, actually, I like the only thing I was thinking of is this. Is this shows kind of the?

A

um This is like a great like looks like a great product, and it shows like the wide sort of variety of use cases of use cases that could be achieved with particular technology, and it's like I, I mean I like that a lot because, like usually, we think about something you know that uh you see you kind of like with your messengers or like something connected to uh like the dows and stuff, and this is kind of like the whole new end goal of thinking about peer-to-peer.

A

So I love the concept. I I hope that you will continue working on this, and uh fluence would be like the the great fundamental for for this product.

F

Yes, I I definitely plan to continue and I'd like to see how it might implement other or impact other projects across the uh web. 3 space.

A

A

So anyone anyone have any questions, comments, uh ideas.

C

I have one question for morgan uh and it's about fluence and it it's about like. Why does this uh fluence network is uh uh the right solution for you for you now for this for your project, like you probably been looking, I I had a look at your um get up today because we had a communication, and I saw you you experimented also with other, like networks. So why did you choose to work with the fluence now.

F

Yeah, so I was, uh I guess, this kind of started, maybe two years ago just experimenting with these off. Maybe you might call it art, I'm not sure concepts. uh I started using something in the domain of hyperswarm or hypercores as sort of a thinking about data as like, if I'm owning my own data, it's like local storage to my own device, so I have local storage and I have access to global storage, something like ipfs, then, when I heard about fluence, I really like the fact that I could.

F

I could adjust the network topology in an easy to use way with their additional virtual machine, so that was kind of attractive to me and the fact that it's a side car to normal ifs. So that was a big benefit. I could you know storing data uh within these scripts. uh I can retrieve them and I can do like a compute value uh in a decentralized way and that was really attractive.

F

If I'm doing like, let's say some type of data flow counter for uh to say like if I'm you know meditating with this data flow in a peer-to-peer way, I can.

F

I can compute how much I can compute how much uh like for my example is basically when I get to a per kilobyte or per megabyte, how many trees can we plan if I'm donating data and have the connection to actual smart contracts in let's say ethereum blockchain, so that was um that was sort of other considerations and then um the final project that I might be exploring the future. I just I need to write my head around it.

F

It's possibly using urban's network as a communication protocol, but then that has limit some limitations and I mean I'm, I'm no expert. So I'm really just a curious developer wanting to learn these different tools and so far it meets uh the standard that I'm looking for. um But I'd be curious to know if there was things that you saw or things that you might recommend that I that I look into.

C

F

C

Yeah, basically,.

F

Topology and connection to blockchain or kind of big ones, I'd be fast.

C

Would it be fair to say that fluency is in a way very programmable, because we we have very different use cases, and- and I found it also interesting because it it it's certainly like programmable, in a way that I can. I can just do it with it. What I want to do- and that's maybe a compliment to the team- uh that it's it's rather open and um a very different sort of people can come for different use cases and and try to see uh um uh if it can work.

F

Yeah- and I think that uh there was some statement saying that it's non turing complete, if I heard correctly, is that.

F

Is that true, can someone from a team maybe comment.

E

uh Are you asking about not during completeness of ear yeah?

E

uh Well, it slightly moves towards being tearing complete, uh but we haven't checked so it was designed uh as not steering complete but as it evolves, uh it might finally be during complete. So I'm I'm not sure to be honest.

F

Okay, no problem, I just I I just I'm, I'm again, no expert, I'm just sort of full stack developer here, but I know that uh you know. Questions of our cognition are are very, uh very broad. So it's like maybe mapping the development of fluence alongside the development of types of neuroscience might be kind of an interesting path. But I won't comment.

A

J

Morgan, can I ask: um how did you turn into web3 space like what triggered you just just out of.

F

Curiosity uh yeah, I guess like uh I, was really into the culture of uh the hackathons, uh whether in person or uh remote, um that was. That was a big one for me and the impact that I had on some of the global values of um you know implementing, let's say, financial transactions in a in a open, transparent or even privacy. Driven way, some of these values are really really true to me.

F

um You know censorship can be an issue here and there, but I find that some of the some of the goals that at least I'm focusing on right now as related to sustainability and some of the uh problems in the medical domain are uh really fascinating.

F

um But definitely what what drove me into the space was uh the ethereum, hackathons uh and sort of the benefits that we get from having this sort of global global, open financial network, and you know removing the middleman- can be a huge, uh huge positive with creating the types of efficiencies and reallocation of resources that we might strive for. uh For you know the type of abundance we seek.

F

So it's not really that direct of a question, but I guess it's been a long bit of a path, but I came more from the more from the centralized corporate domain um building sort of off uh mvp projects and uh I just find that the fast iteration experimentation to be really attractive from a personal development perspective and then from a global perspective. Those um those other goals that I mentioned.

J

Cool. Thank you so much. That's that's very interesting.

E

A

Also on fluence value, fluence is super available because it allows to compose all the things together like if you would want to use uh like to combine off-chain per-to-peer protocol of organ with the protocol of uh euro and then it's possible to do on misfluence, because it's kind of designed this way.

A

um So this is this- is the the one of the biggest value that uh people not yet fully understand, because there is not much things yet blue confluence, but it will be more and more clear. It's kind of same composability that you have on chain but of chain in a particular protocols which is cool.

C

Maybe the word I was looking for before was like unopinionated in the sense that that do not think that much of a use case, or at least that I'm not aware of it, which is a good thing, because it just attracts uh people that want to imagine their own use cases um and yeah.

A

Yeah yeah yeah yeah, so basically this.

C

Is the same sense why ethereum is uh uh uh becoming the biggest blockchain because they didn't think of what it should have been, uh how people should use it.

A

Yeah exactly I like, I love this point.

A

I love this point and uh I think it's uh basically the the the power of com, this the power of community that explores all set of use cases is much higher than like. You know when the team, like the founding team design, decides okay like this will be. You know the blockchain for this or the blockchain for that. uh So this this great approach we are. I think we want to sort of try to repeat this with fluence to be unopinated an opinion.

A

Oh sorry, unopinionated uh as much as uh as we can and allow people to you know, be creative and build very different kind of things, and then figure out, composition of them and so on so forth, because it's all kind of um doesn't depend on any central entity or notes or like a global consensus, or something like that.

A

All right um should we wrap up, then. um Thank you. Everyone thank you, valerie for the great presentation and the deep dive to marine. That was amazing. uh Thank you, morgan for this very unusual product presentation, so very nice. I love it yeah. Thank you. Thank you. Everyone.

J

Thank you morgan. Thank you. Everyone.

A

All right see you in two weeks, bye, bye, see you.

J