Fluence Labs Community Calls, 29 Jun 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: Developer Community Call #24

Description

Today is the Developer Community Call day ✨

🗓️ June 29th at 6:00 pm CET / 11 pm CT

Here's what we've got planned:
• Progress highlight from the core tech team;
• Upcoming plans & events;
• Q&A session

Mark your calendars and join us today!

We meet in zoom 🔗 https://us06web.zoom.us/j/82025547767

A

Let's wait a couple of minutes or everyone to.

B

A

It was a long time since we had our last Community call.

A

So well, I was still joining guys. How are you doing.

A

I was just how was your day.

B

A

Great great, you see thumbs up from Bernard Mike.

A

Okay, so let's wait for another minutes and start our community call.

A

Okay, so I guess it's uh it's already time to start so again, hello, everyone! It was so nice to see you here. uh This is influence, developer, Community, call number 24. and let's see what we have on our agenda.

A

So today uh we will talk about the proof of capacity update. We will be rolling out um by the mainnet launch. We will also uh talk about progress, highlights from our core Tech Team uh and have the Q a session with them. uh We also share our consensus and computer over data. Summit highlights uh and have the Q A's on them and also share, like our future events and of course hand out our co-apps right. So um first things, first um uh Guinea is going to tell us about the proof of capacity Guinea, hi, hello,.

C

Hello yeah: can you just uh switch to the next slide? Yeah uh I, like I, can I would probably like give very short introduction. Just to remind her to everyone uh like a big part of women's project is creating the marketplace of compute providers right because um we don't run servers ourselves. We are not centralized company, uh we're not trying to replace clouds with another company that that runs servers, so we create a protocol and the marketplace.

C

So any one of compute providers can join the network can be Provider, Host users, applications functions and earn money.

C

um So as a part of protocol and as a part of marketplace, uh we have um basically what we call like proofs proofs of either processing or execution or capacity.

C

uh Proof of processing and execution are related to um basically security of computations that happen in influence and verifiability of communications that happen in liquids like whenever customers developers deploy their applications, their functions, um two providers, basically, as as a part of executing those applications, providers, have to um create and submit proofs that are verified on chain, and you know the designs of these things are still not yet fully finalized but processing related to Aqua. It's basically about ability to verify uh certain particles that were issued by applications.

C

Proof execution were designed were named um as a name for breathability of marine function calls. So, basically, when our particle is verified, it could be verified very different ways. It could be like audited by like whether particle corresponds, the script, uh that's in their IR script or, for example, air script plus, you know peers, that's supposed to execute it or airscript, plus peers, plus uh service calls results.

C

So the particular definition of this proves is still kind of work in progress, but that's the the scope of um on this book.

C

So this proves cover the security of computational influence, Network and the whole idea is that you don't need to trust the provider that providers do their work correctly, like you have this probabilistic uh verification that that allows you to use um this network in a trustless fashion and then another big thing is uh proof capacity and what is proof capacity is basically the idea to incentivize providers to bring compute capacity to a network, even if there is not yet enough um demand from customers who want to use all this capacity.

C

So it's basically, uh if you further, if you switch to the next slide yeah, so basically Pro capacity is some sort of work that the providers have to perform constantly to prove that they that they allocating resources to a network at the moment. So they have to put some collaterals. They have to lock some cholesterol. They have to promise that they will be um serving like.

C

You know some CPU memory uh resource to a network over a particular uh period of time and then during this this period of time they have to constantly say yeah, I'm, I'm, still um I'm still allocating these resources performance, Network, there's they're not allocated to something else, so I haven't switched them to something else. So I have I need to constantly prove um that I still I'm still allocating these resources. And this all we call proof capacity and uh uh we consider like we consider different ways to approach this method.

C

But basically, it's a kind of proof work a little bit similar to proof of work um and uh the whole idea that it should be Asic resistance because, like we don't want Asics to be on a network instead of real servers that ready to switch to serve uh user demand, um and uh it should be like this mechanism should include uh token collaterals, because we are interested in um providers to commit uh providing resources for a long period of time.

C

So they have to put some stake up front um and they should earn some rewards for providing these capacity. So it's economically makes sense for them to do it, and this approach is actually like.

C

um You know, used in in Falcon Network, for example, and falcoin basically rewards storage providers to to provide um storage capacity to a network, even if it's not used at the moment by useful data and then gradually this capacity migrates to serving useful data, and um this is a very similar logic. A lot of these things are still work in progress. Like a lot of details.

C

That's why we're not publishing any kind of documents on it yet, um but the we consider this as a as a important part of the my net launch, so um basically influence mainnet should um start from um incentivizing the compute providers to join to like expand the basically supply side, the the um compute capacity side and then start gradually onboarding. The customers from the like from developers to this Marketplace um yeah I. Think that's uh rough. Just rough overview on on this.

C

uh It's smallest hurricane on my side and uh I think we would have some discussion later related uh some pieces of it right because we're going to talk about signatures uh that related to processing and um but I'm, not sure if you have anything on capacity on agenda.

C

Yeah, but that's roughly, like the ref over you for everyone too, to know what we are like: what are some of the directions we are working towards.

A

Hey, thank you so much um so uh yeah right now we're it's! It's part where we discuss the highlights from our core Tech Team, so I can give you the brief overview of what we're going to talk about. So first of all, um we have released uh an edit support for uh Docker images with uh rm64 architecture. uh It means that uh right now developers can use m1mm and M2 Max and gets like the native binaries for them.

A

um So we're also we're also releasing part particle signatures and uh and Mike we're gonna tell about this a little bit more shortly and we're also introducing some new updates into fluent CLI and we are going to sunsets uh Aqua CLI, so we have transferred most of the functionality into fluency CLI and uh we're going to stop supporting aquacy like uh in future releases. So right now, I'm going to pass the mic to Mike.

A

So he's going to tell us about particle signatures.

D

A

D

Either Guinea said so, we are working now on a proof of part scene so, like basically now we almost design how it should be, and we have some missing pieces like that needed for this proof of person, and one of them is a signatures uh or in other words as proof. Caring.

E

D

uh But before like we're going to talk about proof, camera data, let me briefly explain what we're doing we can like kind of More Level terms and uh how, like our network, is structured and how we want to actually submit this proof of arts and to the on-chain part. So on this side you could see like this kind of scheme, there's several Aqua VMS, so Aqua VMS is a interpreter, so don't operates like air code, so here is resulted from uh Aqua compilation of aqua.

D

And uh could you please subscribe to the next swipe and actually it's important that every peer influence Network contains where it costs this uh Aquarium uh and the particle so there's another conception is a particle particle is a you could consider it as a network packet. That's uh like going from one uh peer to another yep, and this particle is basically handled by Aqua VMS. So, like all peers all um run times, all of this is kinda controlled or ruled by aquarium.

D

So aquarium tells how exactly Network should behave, and could you therapist fight to the next fight? ah Yes, so, and uh regarding proof of practicing, uh basically we'll have the division to two parts of chain and on chain. So now we're thinking that on chain party will be on the file coin, and uh on this side you could see how like a principal scheme of how uh like it could be it.

D

Could it could work so, basically, um uh every peer or every Aqua VM on peer mines, a particle Yeah by mines, I'm by mining, I mean that every period after, like a new article created, it checks it hash uh against some, like some another hash, yeah like in a general proof of work. You know this hash satisfies as complexity or like an another hash. It doesn't matter, then the spear like mind the particular hash and can submit it to on-chain part, not not just hash, but the data.

D

That's uh verifies that this, like uh execution, was done correctly here that this was actually done uh and on this right, you could see. That's like this peer one of the of the three peers uh submits uh so-called golden Network packet or golden article. uh This proves uh some proofs to launch game uh and it's important that every peer that's participated in the form. You know this article uh or this like in a particular execution or participated in this particular particle execution.

D

uh They will be rewarded so that for every period it's important to kind of check. Particles against the complexity and submitters are kind of satisfied as complexity, and also it's uh important to send it next to next peers yeah, because, like all next peers, have the same probability to mind as golden particle.

D

So that's how it should work uh like in general, but there are some Mission pieces in the scheme and further, could you please move to the next slide? Yeah, and uh basically there are some requirements that we haven't met before, like first of all, peers must not be able to Tamper data. So it's uh connected to the last uh kind of last point to this list is peer.

D

Peers must not be able to grind it here so like if Pierre uh somehow like I, can get heads of data, and then it should be satisfied to some complexity. Then uh this this peers, they can grind it. They can could like iteratively change, something and data, and, uh like uh just do this, mining, not useful, not useful work and also they can hamper uh like, for example, call results of a cows that was done that were done on previous Piers.

D

Like uh appears that, like before this execution also uh on chain wire, should should be able to verify particle according to air scripts. Here they should be able to check these proofs according to submitted particle, and also like for uh having this like ability to uh uh having like Computing proof of work, Computing hash, being the reliable source of entropy yeah. So now, at the moment like before, like uh last, the last updates we haven't had this, but now with this proof, caring data, you can use signatures uh as a source of entropy Hillary.

D

Could you please go to the next slide yep and a couple thought about signatures and about how aquavium Works internally, so basically aquarium has no any saved entry point so and when, uh like each start of that execution, when particle came to, aquavium execution starts from the very beginning, so it means it's. A VM starts from the like the first instruction and checks like if it was already executed, doesn't check that the result are correct.

D

If not, it tries to execute it yep and to prevent the execution of the same calls like, for example, cows or any other instruction twice execution. Trace uh uh is being analyzed and the serialized in some form. On this slide, you could see the previous incarnation of this form. Basically, we had the four states so like only for four instructions, it's par cow fold and up. uh We had States in this trace and on the right side on these two pictures on the right side, you could see that how they structured for par and cow.

D

It doesn't matter how exactly, but you see, maybe.

B

D

Points here or some uh heuristics how? What what are they basically and after this? Why is this changes so further? Could you please go to.

F

D

Fight we highly refactored our trace and now it's heavily Shady, based on on the picture on the right side, you see how, like all this link together, it's rather complex and we won't kind of take. uh You know, spend much time on that. If you have any question, please ask on a q a session, but this scheme is important for poor care and data. For example, these cids are used for uh signatures. Okay, so basically after like particle execution will never appear. Aqua VM uh signs all necessary data and data signed by CID.

D

So basically we combine all cids sort them and then sign all this like uh final list and then uh before and even while execution execution equivalent like every next peer. We check that all signatures are corrected are correct, yeah and no data was tempered, uh so these signatures are quite important also for sellers of entropies.

D

That I already said yeah, because it's a natural actually source of entropy, because uh it's science with a secret peer keys and- and it's not so easy to like kind of grind particles in this way hash of particles in this way. So, basically now we are we're kind of trying to release the network.

D

It's almost done in aqua VM, so by almost I mean that we have like a final final pull request: that's not yet merged what it was made, and now we are also simultaneously working on a releasing it to the network uh and hope that next week, we'll finally have it on the network.

D

So it's basically all for signatures of proof, care and data. So if you have any question, please ask.

A

Frank, thank you so much um I guess that's! uh We can have like a q a session after, like the whole part about the team, Tech updates um are Tom. uh Would you like to tell us about the uh uh CLI updates.

G

uh Sure, hello, uh yeah thanks share my screen.

G

And you see it.

C

G

Okay, so I will tell you a little bit about um the recent updates for fluency line, what we did uh last month, uh so one of the things is checking for updates uh this. This is a feature it's just for CLI to check for updates.

G

uh You will still have to install uh the CLI yourself, but at least you will know that new new version came out. uh There is also a new property in fluency ammo that lets you pin a project to a particular hcli version. It will ensure that all the versions of every component are compatible with the exact setup of your project. You can also check the version of the raspier that is used with your project, so this is just the way if you want to make sure that your project works in the future.

G

Even if the network updates.

G

uh There is also an old build flag for deployment and for some other commands is just allows you to not build the project uh when you deploy, because by default it is built before deploying, and this is just allows you to do some uh stuff. If you want to, for example, sign was module before deployment.

G

uh We also did a little bit of testing, and now uh our core commands and core workflows with the CLI are tested and working in CI. So it makes everything more stable now, because we can check that everything works on each commit and that's very good.

G

There is a small feature for sports immedioml extensions, which were previously not supported. Also, we did autocommit influencia repo, that is just for developers, so it is easier to contribute for them because they GitHub actions bought. Does all the information LinkedIn and generating logs for you and speaking about docs. We found a better library for conflictox Generation.

G

So now, if you want to learn about conflicts, the docs are now a bit better, but still there is a room for improvement and, as Federer told you previously we're trying to Sunset Aqua CLI and it was removed and now Aqua API is used in influencer, like code.

G

It enabled us this feature that people complained about uh previously. We supported only node 16 version. Now it's uh LTS version 18.. So please update in order to use the CLI.

G

Also, we added a tracing flag for some of the commands in the CLI and.

G

What it, what this allows is it brings all logs from Aqua function, calls as they happen. So, basically, when Com, when Aqua is compiled, uh special calls back to your own beer that is inside, win, CLI are made and uh you can see the logs or functions happening in like in real time.

G

So we added airbt if I into a CLI. um This is a command that prints their scripts and human, readable, python-like representation, it's not executable, but it can be used to easily read and understand air code.

G

You can also use the B flag when doing fluence run to display a beautified air script, and so you can make sure what exactly you are running when you do fluence run.

G

There is also a quick start template information line. It's just streamlines the user Journey like our main journey of deep line and Ryan what you deployed, so you can just init and do a deploy right away and run the default service, which is like a part of this template. It's just Hello World Service service and the most recent feature.

G

People now I downloaded this pre-built binaries. Previously they were built from The Source Code by CLI, and it took a lot of time and now it this process is a lot faster. So please check it out. uh There are also lots of other fixes. No improvements and refactorings.

G

And that's pretty much it! Thank you.

A

John, thank you so much. Thank you. So much for sharing the updates obviously live with us. um Any questions for Tom and Mike.

H

G

It looks a little bit like python.

G

Find some something to show.

G

Sorry for that, let me check if there is some a core function, but let's do hello world and check out how this looks.

G

um Provides some arguments so.

G

Yeah, that's basically how it looks like it's a little bit more readable if.

F

G

Like familiar with air and how it looks like this looks a little bit different, but arguably it's more readable and easier to understand. Okay,.

H

Well, yeah I never looked at it. So um that's something already.

G

Okay, that's that's what is actually running when you do aquaran running on your Aqua code. uh It is compiled into this air representation and to easier understand the.

H

Error code like a few line breaks so that the try and then the line break to the call thing and everything against one line.

G

Yeah yeah, the actual air, looks a little bit different. We can like display how the air looks like with the print air. So this is the error code, and this is the beautified version.

H

Yeah I was hoping for colors, but you know, then we could still be colors yeah.

G

I I also asked the machines team about the colors, but currently there is no plans for that. um Fortunately,.

H

Hey the uh the fixed CLI version and the updates, and then you said something about that. It would like um uh also check on the raspberry version or, let's.

G

Say it it does, it will not check uh for raster version, but if you, if you use, uh uh are you asking about this right.

H

uh Yeah, so if I'm fixing a particular CLI version and.

G

Yes, yes, if you've been, if.

H

You know no um so probably I have an aqua version related to my CLI version and I'm like doing nothing for a year and then all the notes in the network have a higher Aqua version.

G

H

Yes, yes, will it and say, like uh sorry,.

G

uh I understand uh not not that at Hope, but if you pin the version of silai, you effectively uh allow this project to be used only with this version of CLI, and if you use a particular silai version, you can do fluent dependency, V A version command, and here you will be able to see all the components and all of their versions, including the rest Peter version that was used together with this project at that moment that you used it.

G

So if you want wanted to run your project, you can spin up your own, uh like uh part of network like in Docker compose with this version and check out your project and how it was working. uh It's it only maybe like useful.

H

In this regard, yeah the local note: it's not the um yeah. It's not checking your remotes once yeah.

G

Not as good as you would hope,.

H

Well, there is lots of time to uh you know: I, don't know, maybe it's not even useful. My part um yeah uh Mike, sorry I'm gonna, review everything you said and then try to understand it at another time because today was not working.

D

H

D

Can share with you our proposal if you want.

B

D

I can share so like everything that I have said uh say ever since they're written in a in so-called proposal. So we have internals.

H

D

I can share share it with you. If you want okay.

H

And I'll be probably in in a cup in two three weeks: I will have more time again. Okay,.

H

E

My questions excuse me as, of course, support it's not about plans. It's about uh Library support, because um the library that I tried before it doesn't support uh color it output when compiled to awesome and uh the beautifier in a utility.

E

Sorry in fluid infertility, it is built as a was model, so if we will be able to find.

F

E

A library that produces calorie support for awesome income for console investment, uh it would be possible to produce calorite output.

H

Well, I, don't think it has a very high price, although that's me saying it anyway: I'm setting up.

A

Okay Yura. Thank you for your questions. I! Don't thanks again for showing us the air, purifier, um okay! So now it's time to take a look at our participation in the consensus.

F

Yeah yeah, it was a while ago, but still we didn't share any highlights from consensus, but yeah we've been yeah in the end of April, we've been to consensus, it was pretty intense week. We had a lot of um networking.

F

We had a lot of talks and participated in the main event of consensus having a boost there and that a lot of file coin uh events- one of them are file called Network base and also disco sub, which is like the centralized compute Community, sharing our sharing their thoughts on everything what's going on in the space, so that was pretty and pretty good for us and also Tom had two talks which was on our YouTube right now.

F

So you can check this out and and learn more about the recent um yeah updates about the project from these videos as well. uh After consensus, we had uh participated in compute over data Summit, which was not not as huge as consensus, but still very productive for us.

F

um Actually, we don't have any pictures from there, but we have Bernard stocks, which is also you can find at our YouTube channel yeah um yeah. We can go to the next slide and and of course we had our internal Summit in Belgrade, which was very, very intense. We spent the whole week all together in planning and working a lot of strategy, strategic sessions and a lot of brainstorming and yeah playing our main net and the future roadmap as well.

F

Yes, um that was pretty good and for the future events we're planning to go to ECC Paris in July, also yeah, to Japan in Berlin, and to talk in 2049 in Singapore. Also, we probably will be in Istanbul for the deaf connect week, and maybe something else, uh but this is the closest one and Paris is coming pretty pretty soon. uh Let's go to the next slide. There is a really really brief agenda of what we are planning, uh so we're gonna, be at the main event of ECC.

F

We're gonna have our booster for the first two days um and also Dimitri probably will come and give a talk. So, if you're interested and if you will be around, please join us there at the mainland and, of course, at a lot of side events, uh we are mostly concentrated on cloud coin.

F

Events so yeah check us out at Falcon network based and Launchpad for they're gonna participate in Launchpad activities cryptocon day, and many more uh also we're going to organize our decentralized compute assignment and I believe we're gonna have more information next week at our Twitter, so stay tuned, I believe it's gonna be huge this time and as for, if Global barriers, this is a pretty famous hackathon.

F

um We are not like really deeply participate there, but you have an opportunity to go uh there with the scholarship. Can we go to the next slide uh so the same as we did uh for if Denver we're also doing for each Global Paris, together with developer Dao community so um yeah? So anyone who would like to go to Paris can apply uh for the scholarship, so you need to is to have a team and um ability to build on top of our project uh so yeah.

F

If you are able to go, if you can scan QR code and apply uh you'll, be able to go to Paris for free and yeah and win the price. uh That's pretty much. It I believe we're gonna share um more about the upcoming events during next Community course. I think it's too early to share the plans for Berlin and for Singapore. Yet that's it probably from my side, yeah.

F

So yeah we also have watch today. I think I'll send everything over later a little bit yeah. Let me just mean the leads.

A

Okay, another thank you so much um we'll be waiting for that Co-op link, um so guys it looks like we have ran out of content for our community call. So thank you for attending. Thank you for joining um any questions. While we still have to have time.

A

Well, um if there are no questions, then let's conclude our calls again. Thank you thanks you for joining. Thank you for uh your participation for your questions and see you on the next Community course guys. Thank you bye, but.