- Follow up discussion from last months sig-clients monthly meeting.
- Spheron will be giving a demo on the all the work that they have been doing.
- Progress updates from Akash Client partners on new features, etc.
- Date: Wednesday, February 15, 2023
- Time: 10:30 AM PT (Pacific Time)
- Recording
- Transcript
- Andrew Gnatyuk
- Anil Murty
- Benjamin B
- Deval Patel
- Dimokus
- Greg Osuri
- Jigar Patel
- Joao Luna
- Joseph Tary
- Luca Fortes
- Max
- Pard Sahota
- Prashant Maurya
- Rodrigo Rochin
- Tyler Wright
- Anil called out all of the clients in the sig-clients Repo. The goal is have a rep from each client team, and each team can give an update on what they are working on.
- Anil started by giving an update on Akash Console.
- Akash Javascript (AkashJS) library has been open sourced. Anil opened an issue for adding more examples to the library. Members of the community have already added to that issue.
- Greg asked Luna what Java client library is being built for?
- Luna mentioned that GO Library is the priority. He is still having issues with GRPC.
- Greg shared some thoughts on Fly.io, a tool he really likes.
- Greg wanted to note the balance between what the user wants, and what the product can offer. Greg encourage clients to maximize for what the user wants.
- Anil asked if Praetor should be apart of the clients special interest group.
- Anil mentioned that another client partner may be giving a demo at the next sig-clients meeting. Stay tuned in in discord.
- This has been beta for months. There is an akash-console discord channel for discussion.
- User auth has been removed, so anyone can try it out without major issues.
- There are issues in the console repo now. There are features and bugs in the repo that the team is looking for community contributions for.
- "Beta" will be left on, as the team continues to improve the product.
- There are a lot of different ways where community can contribute, including authz and other issues.
- Greg asked what a great first issue would be. Anil shared console issue #30
- Greg is tagging "good first issue" for Console issues.
- Luna and his team are still having issues with GRPC. Luna is talking with Artur about how to solve this problem.
- Greg mentioned that the priorities need to be a GO library, Python Library, and a Ruby Library. Ruby library will be needed for some work Greg is doing.
- Java is necessary for Luna's team.
- Max believes "C# > Java"
- Ben mentioned "C# could be used by Unity projects for multiplayer servers. This can be huge given that enough integrations can be made."
- Greg added a link to the chat on how to publish a transaction: https://github.com/ovrclk/eve/blob/main/cmd/deploy2.go
- Greg has been looking at tools and their user experiences. He shared a presentation about workflows at Fly.io
- Greg shared Fly.io as an example.
- Greg noticed that the time to code is very quick with Fly.io. Akash can take notes from this flow.
- Greg also noticed that the failure mechanisms are clean. The Fly.io gives multiple chances to input imformation, as to avoid failures. Some items like naming can also be automated to keep items from failing.
- Greg suggested client teams to look at the examples
- Ben asked how Greg sees these webflows can be used with Akash. Greg says that a much cleaner CLI, build management with buildpacks, something non custodial outside of Github, or JWT could be potential implementation. Command lines for different experiences could also be great.
- Prashant, from Spheron, gave a demo and presenation on the work he and his team are working on.
- The Spheron team is working on a CLI tool as well that they are using.
- Spheron wants to build an accessibility layer on top of Akash. Spheron has been working with Akash for about a year.
- Spheron is using a community pool model for managing transactions. This is an ongoing issue being worked on.
- Spheron team built a reverse proxy on top of Akash. Right now it is just SSL, but there are many future uses.
- Luna asked for clarification on what the value of the reverse proxy is. HTTPs is a concern for Luna.
- Spheron user can add team members without giving away keys. This is a big use case for Spheron team.
- Spheron supports multiple clusters including Kyve, IPFS, Drupal, and many others. This is an area of focus for the Spheron team.
- Users can attach their own domain to a deployment.
- Ben mentioned that the messaging could be improved to let people know that some AKT can be refunded.
- Max asked how Spheron chooses the provider. It does not seem like a region or specific attributes are made specific by the deployer
- Spheron currently chooses for the user right now, based upon providers that the Spheron team can trust.
- Luna asked how Spheron is handling the provider.
- Next release will include dollar price for deployments. This will cater to web2 audience.
- Community is more than welcome to contribute to AkashJS repo.
- Client libraries working group should consider adding spec for Ruby, based upon discussion from Greg and Luna.
- C# can be replaced with Ruby on the priority list for language libraries.
This editable transcript was computer generated and might contain errors. People can also change the text after it was created.
Tyler Wright: Perfect.
Tyler Wright: All right, again, welcome to the second sick clients meeting. I will be taking notes in this tab right here so you all can follow along. But again, if anybody has to share screen or wants to take over there more than welcome, so I'll pass off to Anil to get us started with the second SIG clients meeting.
Anil Murty: Thanks Tyler. Hey everyone.
Greg Osuri: Good.
Anil Murty: So yeah, what I was hoping we could do for this meeting is, I updated the SIG page, the main thing page to call out all the clients that I'm aware of that we're working on right now or the communities working on right now. So I was hoping is on each of these meetings going forward, we can depending on, you know, who's present in the meetings. If there's at least one representative from each of the teams working on these clients, we can divide up the whole time into approximately equal chunks and talk about each of the clients. And then if people have general things, they want to talk about, that are not clients, specific, we can do those as well. Does that sound like an okay plan? Cool. All right.
Greg Osuri: Yeah. Okay.
Anil Murty: Cool. So Greg, you said you wanted to start off with some general discussions about clients. You just I thought I heard you say that.
Greg Osuri: Me. Right, I had more specific. Not General.
Anil Murty: Got it. Cool. I guess since we're still waiting on somebody from sphere on to join, I was hoping that swear on would join this call and show us what they've been working on. I'm not sure if they'll make it. I think them a couple of times. What I can do is I could start off with console and then I can see if the Luna or Max or somebody from one of the other clients teams want it continue on. So just to give a general update on Console console has been in Akash council has been in beta for a couple of months. Now, we have a whole Crash Console channel that you know is open to beta participants We are pretty close to getting this to be to making a generally available and what that means is, basically today the URL that you access console from is the beta or the Demo Dash console dota Cash Dot network. What we're going to be doing is we're going to be changing that to just consoled or akash.network.
Anil Murty: We've already made that change. And then the access to console up until this point was behind the user Roth, and the user credentials were shared on that on that Akash Console beta channel. So, that user Roth has been removed as well. So I'll share the URL to that link right after this meeting…
Greg Osuri: Here.
Anil Murty: but essentially that'll be the GA URL that that we're going to go out with and it'll be open to anyone that wants to try it out or wants to use it. We think the result, pretty much all the blockers that we had on getting this out to GA. Hopefully there's going to be no surprises. You know, that I found that we haven't found yet. There are a few things that a few wishes that are open in the console repository right now. And these are all things that, you know, we collectively decided were not blockers for gaing the product and what we're going to be doing is working with the community to see if people want to contribute and disclose out those issues as we go forward as well, these issues are combination of bugs as well as features.
Greg Osuri: But that's great.
Anil Murty: So good.
Greg Osuri: So Council Akash networks. It's not it's not officially GA but it is available.
Joseph Tary: Yeah, sort of an open data.
Greg Osuri: Open better now. Nobody better.
Anil Murty: Yep. And yeah, so it's gonna be we have with not, you can just get straight to the page and we'll still leave the Beta sticker on there because, you know, we don't have a extensive QA team as such and…
Greg Osuri: so,
Anil Murty: I'm sure there'll be things that we found. And in general, you've got a, you know, Work with the community and other users to figure out product marketplace and stuff over time. So it makes sense to leave the beta on, I mean, Gmail was in beta for, I don't know, 10 years or something like that. So
Greg Osuri: Salon is technically in better.
Anil Murty: Yep. So, so that's the updates on console. I don't think it makes sense to show demo. I think most people have on this call have seen the thing from the Akash console, it is called Channel. But once again,…
Greg Osuri: What?
Anil Murty: if you if you're interested, I'll post a URL up to the GA version of console right after this meeting. And then Joe was there anything else you wanted to add or what I said?
Joseph Tary: Not think of everything pretty good.
Anil Murty: Cool. Thanks. Any questions, thoughts?
Greg Osuri: Okay, get it out. Their console is an open beta.
Anil Murty: Yep, we can. Yep. Yeah,…
Greg Osuri: Rock and roll. Let's do it.
00:05:00
Anil Murty: working with Zac when I say no, that is what? Cool. That's it for console.
Greg Osuri: Consulting. This is big achievement. It's been in the work for a while. It's awesome. Now it's on the open. And people can. You know, suggest improvements or even some pull requests. I'm excited. To extend the console. Myself.
Anil Murty: Yep. yep, you yeah, I know like you said Gregor where, I'm hoping that we'll get more community involvement in contributions and console as well. I know, there's like two or three people that were interested in contributing to Front-in code. I don't think any of them are on the call right now. if there's anybody on this call, that's interested in contributing to console. We have, you know, work that is In in terms of like in a gradient in terms of complexity, there's a very simple issues that are cosmetic fixes to the UI here and there and there's more complex issues like working on authorized playing for example. So, there's a lot of different things that people can contribute to.
Greg Osuri: Awesome. Is there any way? I'm looking at the issues list.
Greg Osuri: What would you suggest would be? If first good first issue in terms of complexity that community can like pick up
Anil Murty: see, I'm just gonna go to console repo myself, Issues. Something like something like Help links to the sidebar.
Greg Osuri: I want to create it out.
Anil Murty: Could be a good one.
Greg Osuri: Hyperlinks, good. Bye.
Anil Murty: You know, the ports row looks cute that when we have had trouble reproducing, it happens on some browsers, doesn't happen on someone's but it's like, well, it's a little bit off to the left to the right. I think one of the things that would be a good one. It's a you know if you're able to reproduce it it's a minor. Um, ui fix. and,
Greg Osuri: From label them as good first issue.
Anil Murty: Yes, yes.
Greg Osuri: And anything you think is a good first issue, I want to quit it out and see what we can get.
Joseph Tary: And we should create a label for those ones, then we can kind of tag them as they come along.
Greg Osuri: I did, it's already there. It's good first. Good, first issues the label. Wow.
Joseph Tary: Awesome, cool.
Anil Murty: Yep.
Greg Osuri: Talking about. Awesome.
Anil Murty: Yeah. On the make sense. Yep. And good. Segway into the next thing I wanted to talk about if there's no more things on console. I wanted to bring up that since clients, libraries is technically part of this figure as well. Under Clients Library is obviously Luna and team have. Thing in terms of, you know, speaking out what it'll take to build Java, plant library, as well as a Go client library. I want to add that since the last SIG meeting we have open sourced,…
Greg Osuri: You.
Anil Murty: the Akash JavaScript client library as well, so that's available under Akash Dash, Networks, Slash Akash GS And so, that's basically, you know, it's used by a few people already. In fact, I think Scott from our team uses it for building a support UI as well. So, definitely encourage people to use Java, JavaScript Library, reporting issues that you might run into suggestions for how we can improve it. But one of the issues that I opened there last week, was the an issue for adding more examples to that repository. And since then one of our community contributors has added three examples to it. So that's another good good. One to attack as a first good issue.
Greg Osuri: Here.
Anil Murty: So what I was doing in that repository at least is I was using a label called. I Think it's I think it used needs contributions or needs, let me just really quickly. Oh gosh, yes.
Joao Luna: I think usually there's a label for good first issue. We might look into it and use it.
Anil Murty: Yeah, that's what Greg got created for console. for for Akashi, as I was using a needs contributor label, but I can change that to Good first issue as well.
Anil Murty: But yeah, so that's another good example. I know this. Yeah.
Greg Osuri: What? It's a question. Remember the clients are Like cleinlib.
Anil Murty: Sorry, good. Sorry. Repeat that oh yeah,…
Greg Osuri: Album Moving the Clients, Lib.
Anil Murty: so we have So you mean like the code repo itself?
Greg Osuri: No matter. I thought would just continue please.
Anil Murty: yeah, so the SIG If you look at the SIG for clients it has a subfolder called Clients Libraries and then there is Luna's work as well as akashias mentioned under that So yeah.
00:10:00
Anil Murty: Cool. So yeah. That is the Update. I wanted to give about Akash Gs and any welcoming contributions to To that, as well as any new issues that people find if you utilize it. So yeah,…
Greg Osuri: Okay.
Anil Murty: Consul in Akash. Yes. That's my updates. Happy to pass on anyone else.
Greg Osuri: Thanks.
Greg Osuri: Awesome.
Greg Osuri: From a clients library standpoint.
Anil Murty: Well Luna.
Greg Osuri: Luna. I think from priorities and point, what are we? What do we build in Java Client Library for in front of
Joao Luna: So Java is a requirement from our site because we have Java developers and…
Greg Osuri: Okay.
Joao Luna: we have some of our services in Java. So that's that's completely on our side.
Greg Osuri: Okay.
Joao Luna: We belong, we own The. The effort on on that because we needed some kind of like the priority for the community but it is though is the priority for the community. As far as I know and we are also tackling it. There's still some confusion on on the grpc portion of it. I'm still in contact with guys from Overclock to figure out because from what I see grpc is not working. We are using amino. We can't really connectful nodes using grpc or there's some confusion on that I need to to better understand it. So I'll try to schedule something in and start tackling it and Terraform providers. Also dependent on that
Joao Luna: Regarding the Java one is though. It's more advanced. We have we are following the the footsteps of the cashier is in terms of documentation and the example code, And we already have integrations with the rest API for for the nodes. And now it's weyers wheels and the code is tested as well. We have integration tests and all set up now, it's a matter of figuring out the grpc and the best way to do it and start on that. These are the updates.
Greg Osuri: If you can be more specific about what the grpc issues, I can be more help.
Joao Luna: I'm not really sure yet, but as far as I know, I have in a conversation with Arthur because I was having problems with connecting to grpc. He told me something about grpc is not working At this stage. There were issues in the cosmos SDK and tender means design preventing from using grpc and amino. At the same time for now, grpc is running via amino is using a gateway. So I want to follow up on this and better. Understand what's what's going on?
Greg Osuri: Okay.
Joao Luna: And yeah, so I'm kind of blocked in that sense on that side.
Greg Osuri: Okay. And the reason why I'm interested in, we need from a priority sample, we definitely need to go library for For. You know, several things were more importantly. The telephone, I need a Python library because we're getting very very close with GPUs. And I also need a Ruby library for a secret project I'm working on. This is a project that we're you…
Joao Luna: Okay.
Greg Osuri: we need a Ruby library because we're, you know, iterating quickly using a rails app to understand the requirements from the ML standpoint for context this Akash AI
Greg Osuri: you know, is, you know, essentially verticalized solution to take a cost to Ai ML Developers, right? And kind of issues ML folks have kind of needs them or folks have very different from whatever three needs to look like for obvious reasons. Um and so we're quickly iterating using a rails app. Not really me. Greg Osuri. It's my side project that I'm irriting quickly using a rails app to be able to
Greg Osuri: You know. To understand what the user requirements. Again, this is a throwaway app. It's not a production app but at least lets me move really quickly. And once we understand what the user flows for the requirements are real transition that to a more scalable, you know, go base or whatever back in based application, we will. But for me to iterate quickly, any sort of a clients support, so the Ruby clients support will be great for a cash.
Joao Luna: Okay, Ruby is not in the spec but I guess we can add it. I haven't thought of of it. I
Greg Osuri: I don't see it quite c sharp, and java you say there is some requirements. I don't for me. I can't. Find reasons, why C sharp. We're supporting a Java. I mean still need to Still use cases, but that's my take.
00:15:00
Joao Luna: Yeah, I get that I think we can from the C sharp. I wanted to include well at least one enterprise language. Let's say we can either go to, we can go to which Java and remove C sharp all the way. And and that Ruby. But Java is is important for for my team. We are we are mostly Java developers. I'm the only go developer on the team. A.
Greg Osuri: Got it.
Greg Osuri: Yeah, out preferred Joe over c sharpening day and I don't we don't interact with a lot of C, sharp developers. Job.
Joao Luna: Yeah. But yeah, the goal one is the priority. Hopefully, I can be unblocked. I'll
Joao Luna: I'll try to keep contacting and get a working. Poc. And because by the time I can, I can contact the grpc node with with the collaborary without Cosmos SDK…
Greg Osuri: You take a look…
Joao Luna: because I want to avoid to Cosmo system independence.
Greg Osuri: what I've done in my Eve repo for Overclock Eve.
Joao Luna: No. No, I can. I can have a look then. Did you use the Cosmos SDK?
Greg Osuri: Yes, it worked really well.
Joao Luna: Because we want to avoid the corporate dependency. Of it.
Greg Osuri: Oprah. Why?
Joao Luna: Yeah. well, because we are not really building a CLI, so we want it to be like, a standalone like a simple as possible from the grpc client perspective and not have all the
Greg Osuri: I don't think you need Cobra.
Greg Osuri: Well, so take a look at this code. I'm going to paste it. Here.
Joao Luna: Okay, okay.
Greg Osuri: C Sharp Review Unity project for multiplayer. Servers could be huge given, okay? Fair enough. So that justification which is sharp is game devs. In the, in the chats. Max C. Sharp is greater than Java. I I particularly don't care about both but sure.
Greg Osuri: I'm not an enterprise developer or game day. Cool. So the the link, I posted tells clearly how to do the most advanced functionality which is creating a deployment
Greg Osuri: and it goes through all the different. gnarly steps to to actually, you know, Publish A an event, a transaction.
Joao Luna: Okay, yeah, I'll see this. This is huge help.
Greg Osuri: Thanks.
Tyler Wright: I had Anil.
Anil Murty: Hey yeah, I just wanted to introduce Prashant Maurya, which is joined the Prashant. So Prashant is with the square on team and it's pretty late for him in your time. So thanks for open on. So late your time Prashant and Espeon has been working with with us for a while now integrating Akash into their solution. So would be great to hear from Prashant about swear on in general Know if possible to your demo of Akash as well.
Prashant Maurya: well, here guys, I believe
Greg Osuri: I can barely hear coming in.
Tyler Wright: And they're muted right now, Prashant.
Greg Osuri: Okay.
Greg Osuri: No worries.
Tyler Wright: Sounds good.
Greg Osuri: In the meanwhile, can I present something? Some some quick observations that have Taken well, Prashant's getting ready.
Greg Osuri: as part of my just general research, you know, I learned tools, I try to and you know, look at stuff that I that Who's that? I use and like, And I look at the flows. And I try to take as many notices as much notes as possible. Right here. Have captured something called flight or IU and how many of you know about this company? I've been, I'm going to take over. Tyler from you.
Greg Osuri: Flight or Io is incredibly beautiful. He's like
Greg Osuri: at a, I would say like, I think they're like primary
Greg Osuri: Stack is ruby on rails, slash. Postgres deployment. Highly recommend, check it out. I have a global replicator postgres. They are one of the Like darlings of use experience and people love using them using those tools. They're like, trying to reverse Erica. They're more modern version of Erica but a lot lot better than heroine terms of you know, just usability and, and the kind of stuff they support, that's what like globally, replicated, postgres clusters, they're really good at
00:20:00
Greg Osuri: Okay, can you okay and look at those user experience, right? This is a sign of flow basically. I've got a fire io first thing you see is a Sign up and it's fairly centralized. So take it with, what what you want. But but the floor I think is here very, very important. This is a sign of page. You get full name email password inside of using Github And then, you know, when you sign up using GitHub, you're presented with two options. either clone if a repository,
Greg Osuri: And actually three options, clone the Repositor, and they have like a Web CLI, which other is very, very impressive. And then they have the CLI, which is beautiful right for five steps you can see Brill install Flight, CTL Price CDL, login clone the repo. They give you Example repo. CD to the repo. Fly launch. Now done. And within five steps you get to productivity, you get to understand how the system works. And that's very, very critical here. The number one. point that we had to learn from this user flow is time to product two day call it, right? So the kind of productivity is so low that people can understand the value prop before, you know, as quickly as possible, right? And
Greg Osuri: I thought I was very, very impressed actually, when I tried a bunch and I created a new tailwind rails, demo app just to see what the user experience for a new. You Know, rails app will look like And I was able to I did like cityloft login. It opened up a in a browser session for me. This is how the login from it from the command line to browser looks like you can see there's some token usage here and you know ask me to user. Yeah, and then that's it. Now I have a full command line and next thing I did was Fly launch. Now it detected my source code, it it detected that I have a rails app And it's beautiful here. It's It's employing a pattern that I a coin called fail. Forgiveness.
Greg Osuri: I failed forgiveness is a interactive user pattern where even if I forget to give an option. Like, in this case, I didn't give a name for the application, it's giving me a second chance to choose a name instead of failing, right? So typically commands fail, When they don't work right failures are very very important. but they're also not so great when the user is a human being and not a computer. So failures are very, very important in the user is a computer because You know, only human being but if the user can do something about the failure more often, not people don't know what to do when something fails for the first try. Right. Without looking at the documentation. Here in this case is saying, Hey I forgot to give an app name. Can I choose where a flan is going to generate one for me? Beautiful, you click on it.
Greg Osuri: There was another beautiful thing. It gave me was to choose a region. I didn't take a screenshot of that but it was incredible. Highly recommend you, you try it out. Says Hey, look, It automatically gave me like a list of I guess 10 to 15 regions and it said something like and it It highlighted, San Jose being the closest. You and it, but it gave me the option out before I can figure it and then I created like this. An admin URL from App, you know set a bunch of secrets. Yeah yeah I directed my docker file and just to the bill for me it didn't fully go through because I had a set. a ruby on rail setting for like secret TBS, you know, Ruby on rails fails, otherwise production but I have no reason to believe this won't work fully but the user for so incredible for me to get
Greg Osuri: You know, to understand what this this, you know, application was and I thought it was very, very impressive. I can share that team. So those of you that are building like front ends should look at in a flight or i/o for inspiration. That's all I have share. I'm going to place this presentation link is open to everybody public.
Greg Osuri: Anybody with a link can use it, but it'll be great to see in akash workflow like this. For whoever is building this app Benjamin.
Benjamin B: Yes, I want to ask you what you think would be required to make this extra plate to akash. Like not only like the CLI being translated but how do you reckon that enough examples will be built to support a similar? And better applications.
00:25:00
Greg Osuri: It's beautiful, right? Yeah, you need a few functionality here, right? So unity, a cohesive Federated user experience between CLI and the Web. Which is number one. So I mean to be a tool on top of Akash, right. I think like it's
Greg Osuri: One way I can look at is the web version doing the authentication with a Kepler. Basically instead of using a github login, And the CLI version will route to the Web. Web version for authentication, instead of relying on the You know, on the, on the key set of system. So I don't know, I think there's like one sort of like Challenge that that we could see. But one way I could propose and several ways you can do this. But one way I can think about is Is the CLI completely depending on the Web for authentication? And not have any keys locally so that way, you know, you can use Kepler what not to sign it on the Web. And then create some sort of authentication token, some session token, not sure yet. No we're looking at JWT now for as a potential solution for this problem and then and then the second step would also involve build management. So the the app needs to know
Greg Osuri: You know what? Kind of application you're building here detected Ruby and rails, right? So you can use buildpaths, buildpacks are incredible. Yeah, I'm going to share this. That's going again.
Greg Osuri: I know those. Some of you were exploring Bill packs but cncf Build packs. I'll be back, right. I think is
Greg Osuri: Project. This is a CNCF projectile Cloud Native Foundation project. It's
Greg Osuri: it obviously began with Heroku, bill text, but it's incredible. You can pretty much either build any application you give it. You know, it does, let's see the registry.
Greg Osuri: There are Java app right? Like The shower app you can actually there's Java functions up too. You can actually create function as a service fast on Akash Using this Java functions app. You know this a whole lot of like You know. You built back just for Java, right? If you look at rails, for example.
Greg Osuri: there is rails and then you have a ruby. So many Ruby apps, right? And then you have an JavaScript
Greg Osuri: because you gotta do react. Mmm. Yeah. Okay, look at the actual dirt bike, but it does detect, you know, the JavaScript apps. I did, I did publish a react app using the back. There's tools like this, we can use to do the build management and then you have container registries. I don't know if you can do it fully fully like
Greg Osuri: Non-custodial that'll be that'd be great to have something like the fully nonca studio but Esperan or I want you Guys are in a very very good position to be able to do this level of experience.
Greg Osuri: Yeah, thinking through some of the stuff.
Greg Osuri: I know I didn't answer the question fully but just give you ideas as to where we can explore some of the areas. We can explore are these
Benjamin B: At first, I thought you were not gonna answer the question, but then when you started showing in the middle packs and all the different things it could do to help him build that up. I think it explained the more than enough
Greg Osuri: Thank you. I mean, I took a stab at something like this. I wanted to build with Akash a while back and then CEO job took over from my program, or job. So You know. This project called Eve. I you know, I've been working for a while it's open source. um,
Greg Osuri: It had.
Greg Osuri: Some of the argument for something, this is the same, the user. Should I be punished for not? Providing the proper inputs. I also like, You know.
Greg Osuri: Gave like how the output should be. Like it's human readable. To grab parcel or JSON parts of all in the sense when you're actually having like systems using your command line. So humans right So I think you got a design command lines for different experiences because commandants can be scriptable or usable, right? So by default I think interactive mode should be the the mode where it's users but you know having other supporter be great. Yeah, take a look at it.
00:30:00
Prashant Maurya: Yeah. Hi Greg is. Is my voices are even now.
Greg Osuri: Yes.
Tyler Wright: Yes.
Prashant Maurya: Awesome. Shall I go ahead and present some of things? Okay, I believe whatever you have been presenting. Just now click on the CLI. We have been working on it, there were some frictions, which we have been. Facing. But we are launching First Year Live version. Within most likely. Next month around March around. So looking forward for yeah.
Greg Osuri: Welcome back.
Prashant Maurya: So it's my screen is visible now.
Greg Osuri: It's coming up. Oh beautiful. Okay awesome.
Prashant Maurya: All right. Yeah, so generally so I spit on as you might as well have seen like what we have been doing for to build a accessibility here. On top of the, the infra providers out there. One of our partners like of course with him we have been working. I don't like it's been around more than yeah around a year now understanding. And and trying to figure out like, what exactly the users are going to love once we give them and that is what we have seen. So there were a couple of things which we optimize for our users who are onboarding on, why it's been on. The one thing is like, which we have basically gone through was there, there was, there was one of very straightforward use case, which we were kind of looking forward was around how exactly we can minimize the process of, let's say, allocating the ground to the people who are going to be today or tomorrow.
Prashant Maurya: Um, the problem, which we have also seen, I believe great and more than or most of you will agree. Here is like the giving a granting in in fresh space. Versus giving a granted device space plays a very vital role. In fry space, Grand can directly be utilize into the infra by giving out to the like we should be worrying more about like if you're giving up the grant that should get utilized by the users, no matter how. And that is what we have kind of. But I believe we have been playing around with our liquidity for a longer now so whenever whenever and you use a locks into this around they get by default 2880 from our side. It is free like so and when I say they get by different take it, they really don't get 2880. They get an alliance of 2080 to spend from our liquidity pool and that is where we are kind of creating the liquidity pool model, where community can contribute liquidity, pool and allow other parties to also earn from that liquidity will try to bring somebody into it. Also to like
Prashant Maurya: Read more and more income for this equity will just answer over the time, once they start providing the liquidity into that pool but we are trying to figure it out how we can secure it much more deeply. So that is the reason we have not yet opened up but our framework is said this is what it does.
Prashant Maurya: This is the first model, right? So on the ontake and that that is where we have seen like around in this Farah itself we had spinned up around more than around 400 plus containers. So far out of which 100 like many got fail initial days because a lot of things which we were building and out of which I believe now 60 50 60 Even now I believe Barcelona eat hackathon which is going to be happening. They must be using Sphere on and akash for launching their computer and everything. That is what we are in discussing with in their Dev. We are also in a talk with a lot of multiple companies in here in India and also in Asia and multiple reasons to go ahead and and start. And one more thing is like, one more gaming project is going to be launching soon. Using purely using Akash and it's around completely into an, and we have seen a tremendous growth there and one more thing, I just want to show you guys There was when we were developing, with the gaming project we found on YouTube was around the SSL certification and we identified that and we built a reverse proxy on top of Akash, which
Prashant Maurya: Means, like, If somebody goes to surround today, they don't even have to worry about attaching their akash like SSL on their docker container side. They can just attach it over its side, Just go annotize their domain. If you don't use the by default SSL, they don't even have to worry about anything now. So it reverse proxate and and you get the bandwidth and you get whatever you want from from outside. It's it's up to the like users, how they want to basically take it forward and we don't even constrain them. Now by not allowing to go and use it manually. There are a lot of other ways you can go into it. So that is what we have done so far. Just to give you the quick demo is like one of the one of the interesting thing which we have all seen like the members and the contribution part. I don't know, like in our cash today, is it possible to on board? It's a team member. It's a great has 10 different team members who might be using the entire state farm most likely Greg has to share. He's private keys to all this team members to basically go and access that. But with this channel, you don't have to do any sort of thing. You go and be on any channel support.
00:35:00
Prashant Maurya: Know I do Team members, doesn't matter. The reason we wanted to include this into the computer space as well, because we have seen a tremendous use case. There we are most of our, most of our companies whom we were reaching out. They their CEOs were not, they were very interested to go and talk to but they wanted the seat is also to honorable, but they are initial call was like the CEOs. So we had to basically demote them and on board. And then we have to ask to invite the CEOs. So that is how the invitation mechanism we thought. Like, Okay, so compute is also going to make sense here. So we have also added into it, so it was not earlier there was not working, it was there but it was only So now it is enabled, we people can go and in my team members, it is very easy. To go ahead and do that. Apart from that, all of those things remains the constant there. But there are a few things, which we
Prashant Maurya: improved heavily was around the clusters. So today, experience supports like multiple clusters. If you guys have not checked out, just go ahead and write to go and check it out. We have the sooner, we have, we have, we have partner. So yeah, we mental very highly. I like now stride ipfs and and all of these things that you that you see today and our intent is to and even hyper sign from, I really talk with a couple of more, no providers out there to basically provide user whatever can be done at the easiest fasting at the current stability of the cost. We are trying to basically exploit that I believe we also have the discussion with Jason well only using them going forward for deploying or some of the computers as well. So a lot of things which we have done around the marketplace so marketplace is very smooth and very clear now. So let's say if you have to go and launch your iPad and this is what we are also going to promote down the line. So let's say,
Prashant Maurya: Will have to go and launch your IP address node are. Today, it's very easy and straightforward. 32 gigabyte of SD to CPU and it is going to cost you this much. If you want to go and see the in the dollar price, we are currently working. So next release, you'll see the dollar price on the, on the front line. But so let's say I want the 32 gigahert which is going to cost me around 8.69 per month, I'll select the plan. I'll go and say Continue deploy the moment users device. Now basically we educate people about like what is going to happen. Once you hit on deploy, we tell them There's a five ticket which will get blocked from your account. It is does not means, like, it will be like directly off once you close the inside back. But this is what we just so that they don't panic. Like, we're exactly 81. So just say, Don't continue. And the moment you don't get nice notification here, you'll also get. So there are multiple things which we enabled in the platform, the notification on email side. Yeah. Enable notification on the browser side.
Prashant Maurya: We make sure like we kind of provide all the experiences what user are used to of facing into any of this. And like, it's it ultimately are in police to just push as much compute onto diesel, as network as possible, even though there is something less component, that is what we wanted to do. And that is what we are kind of doing it as we move forward. As us clients decays will kick into the picture and if you see like so now the instance has been deployed. I got my notifications required for Ipface if I refresh my link. I'll be able to get that particular link as well. So, let me just differentiate here once. The instance, instance will be up, you will start getting the like locks here, but it just takes certain time to basically heat up the container and then start showing the logs as well. So these are the couple of things which we have done over the over the couple of months down the line. The next thing which we are going to be publishing, the first thing which you are going to do is around these firons computers. Yeah. Before we jump into that, here is the domain
Prashant Maurya: time where you can go and attach your domain, Whatever, the instance, which we have deployed, you can go ahead and annotize that and subdomains or whatever like Depending on like user what they're kind of looking for. If they're looking for attaching their own domain, they can go ahead and do that if they're not like, they're not. So so that is how exactly it has been built so far. I need to check like where exactly the deploy logs went, but this is what we have done so far. Yeah, Benjamin you have, you raise your hand. Would love to Have your question.
Benjamin B: Yeah, I just wanted to add that. When you click the deploy button, it looked like to my IC. At least like the font would just disappear. I know afterwards is to refundable amount, but I think this should be made. Clear before pressing that button as well. If it isn't already, I might have missed it.
Prashant Maurya: No, come again. What exactly it is.
Benjamin B: When you press the Deploy button like before you went to the page your own right now.
Prashant Maurya: Okay, let me let me do it. Let me do it.
Benjamin B: It looked like it was on the clear that the akit could be refunded. Like, it's
00:40:00
Prashant Maurya: Yeah, that is what I said. Like we'll be we'll be adding that message as well. So we have got that message from multiple users like they said,…
Benjamin B: Oh yeah.
Prashant Maurya: like Hey, you know what? Yeah. So we'll be a kind of completely. So that is what. So, it was our earlier iteration, when we have started showing the users that I think it will be deducted. But as so, let's say, if I go here, right? So if I go to the building, I can see my computers. Now I have 14 earlier, I had 19 Now let's go and let's switch off my this. Particular ipfers. Thing, which got deployed. Let me just go and just click.
Prashant Maurya: The moment I'll delete it back. But yeah I I agree on that point is like sometimes people get panic. Like what happened? Like if it is going to be refunded or not, we will add that message also a specifically and even even not even message. We are kind of moving that a kitty power itself directly because that confuses to the user in terms. Like, so for example, if you go to rip to users, right? They don't even care if it is, if it is, I'll be very honest, right? So they don't care if it is akashi, if it is a sphere on, they only care about one thing is, is my Drupal app or is my workplace app, it's getting deployed on it is working or not. Okay? And and that is where we wanted to like, go ahead and add that and bring that dollar value, instead of a kitty value, it will be there, but it will be depending on user. If they want to fish today, version, they can switch it. If they want to stay on the television, they can. Yeah. Any more questions, please go ahead.
Greg Osuri: Luna, you want to go ahead?
Joao Luna: Yeah. Could you clarify? I got a little bit lost on the On what you mentioned for the reverse proxy. Why was it needed? I could you clarify on that?
Prashant Maurya: Yeah. So let's say, generally what happens if I am running? Okay, so I'll give you the credit class example. Which link was facing the gaming project. Okay? So what they were doing is like, they were running their load API of officer on via their server. So the game basically provides, the itch.com kind of functionality where a gamer can come drag and drop on a browser and that Browser basically pushes that particular data packets to the backend and back and support the data too.
Joao Luna: It.
Prashant Maurya: It's Firons server now is the country running on HTTPS, but when, but the Akash server was running on HTTP, okay? And when they were making a call from HTTPS to HTTP, it was throwing down there and that was again, the problem. The next thing,
Prashant Maurya: And the next thing which we also identified like why we also have to give it is like so even though today also I believe you can directly go to a customer SSL there if I'm not wrong. But but these platforms they were using from us directly. And that is the reason we wanted to add it in between as well with the caching, our main with the caching out. With, with this reverse proxy, Our main intent at the long term is to also provide the Syrian layer on top of it. So let's say if you're building an API layer, right today, there is no way you can go and capture somewhere while you're receiving that, it can never happen. If you, if you do it on a raw metal server, I dare metal server directly either. You have to build the entire
Prashant Maurya: Proxy server on your own and then build it on top of it. What ski shop. And I believe that's much better where you can just attach aPL here. Caching, but that caching does not scales very well. It just stays in the memory side. So let's say If you have to scale originally, then again, it creates the problem. That is where we wanted to introduce that reverse proxy to make that happen. So and then people can basically go and configure on that reverse proxy from which locations, see we all live in. Decent Life is one thing which also affects it has a law of land and I personally believe into it. So a lot of languages, like if the user of this particular region should not be allowed to contact this particular server. The, the user…
Greg Osuri: If?
Prashant Maurya: who is using let's say if a WordPress user is there, he might not want to target the different countries. You might want to target, really one country very easily. You can block that up front into the US, proxy directly instead of configuring it and everything. So that is a long time. So this is why we wanted to just put it there. Luna. If I made the right.
Joao Luna: Okay, okay. Yeah, you did because I was just curious because of the limitation on the HTTP because I've been pushing TLS termination and There is something I'm currently working on, which basically at this, I think my provider is now the only one that has TLS enabled. So, anyone who deploys on my provider has https of the box, I think you can check. If I've done like a website, you can check my personal website, we did that URL. You'll see I have a valid certificate so that limitation I guess we're on will benefit from
Joao Luna: As a casual beyond HTTPS.
Prashant Maurya: Yeah. So but again again it again, it starts,…
Greg Osuri: You.
Prashant Maurya: Yeah again it depends on the so let's say if you have to make more configurations around it so HTTPS is like.
Joao Luna: Mm-hmm.
00:45:00
Prashant Maurya: So let's say configuration was around the regional detections, right? Caching layer and then so a lot of things which can be like built on around the reverse proxy. Now, HTTPS was one of the part because that is what we have faced. And, and Link was like, again, we have to just find out the particular provider. We score providing me at a TLS whenever we are deploying it so let's imagine…
Joao Luna: Yeah.
Prashant Maurya: if any point of time we don't have that like provider he's available, right? Then we are screwed.
Greg Osuri: Yeah. So,…
Prashant Maurya: So that is the reason we just wanted to
Greg Osuri: I mean, we're An interest time we're presentation is going is like right now it's it's a sell but they want to build out futures in terms of data residency, in terms of data access specially with ofac compliance. Right? So now we're seeing a lot of Defaaps, like Mars Protocol wants. All of that compliant. Choices code that into your app or code that into a separate layer. I would prefer code that into a separate layer because app should be pure. So, they're reverse. Proximity, is helpful blocking IP address is really helpful. Caching, My Prashant said is really helpful. You can figure your your Ruby on rails or app to leverage boxing. If you want a lot of capabilities, everyone's proxy will provide you that Prashant. I mean, Surround team has has aspirations to extend to
Prashant Maurya: Yep. See. Absolutely right? And that is where we wanted to fit towards and our next release I believe.
Greg Osuri: Right.
Prashant Maurya: When, once we come over next to next sagic call will be more than happy to. Also show some more progress around how exactly the CLI and the things are looking. Like, we'll show it first the Web hosting site and with that, you guys will be able to figure it out. How exactly is going to look like for computer as well? So it will be very straightforward, like, what you have soon, like slide.io. It is going to be more around similar, but we are kind of working on the on the Central Florida at this point of time until it takes and things starts coming in.
Greg Osuri: I'll be right back. I got to answer the door but continue
Max: Thanks, awesome. Presentation and professional. Just have a question about the way you guys choose the providers and it looks like you well, I wanted to get your well just get your feedback on how Do you guys choose the provider when you deploy, it seems like you don't specify like a region or something or
Max: Any specific attributes or something like that?
Prashant Maurya: Right.
Prashant Maurya: Right. So generally what we do there were couple of things which we have also since we have been working long with the team right over the with Akash. Now, a lot of provider, which we have found, and I believe we also have a private child. Like, we have a chart group with our casting, as well, where Tyler and all of these members, they're from the Cascode core community. So we tell them, like these are the providers which are not working properly. And we also tell them like This is how exactly it is not looking like as of now, we have not built our own reputation system. So what we have done is like, we basically choose from the latest build possible. So let's say somebody is bidding, the lesser amount with choose it for the users, but as I said, like as we move forward, we'll give more and that is where the US proxy and all of this things are going to be coming into. The picture is like, really developers the more and more options to go and choose the regions for the deployment. That is not not at this at a very highest priority for our US at this point over time because we are on more on
Prashant Maurya: Stage where we want to see, like how much interest people are kind of genetic. And for example, so we recently, I believe you also what connectivity server as well. They have been also Onboarded on Akash as a provider. We are also kind of sending some loads on onto their server as we're on the high pride to see like how exactly they are kind of working. And and it just started day before yesterday, it caught integrated. So this is how it the selection process works at this point of time. Going forward. You are going to see like you can basically choose the bits. You can select which video you want to choose will not automate if you if you want to automate. We'll we'll be more than happy to do it. We'll charge for it. But at this point that this is what how it is designed. Max.
Max: Okay, cool. Thanks for the information.
Tyler Wright: Luna, let's ask your question, but
Greg Osuri: Luna.
Joao Luna: on that, I would just It's a really really quick question. It's How are you handling? Like the architecture of the provider? Our new assuming everyone is x86
Joao Luna: Wouldn't tell that your apps break…
Prashant Maurya: yes, this is
Joao Luna: if they are if they go to arm because my provider is arm if someone is unlucky. And I bid on a On him.
Prashant Maurya: so generally I'll tell you tell you what happens is like We have seen this issue happening. So what we have done is like we have developed our own content where where's few of the users who are facing that. Let's say if they were deploying on Linux and that particular provider is getting selected and they're not supporting that Linux one. So we tell use because we have already found those users. So we tell users, Hey, just change it into your docker image on the top, and it will work. And, and we make sure like that particular goes into that. So, this is how we fix that. Keeping that entire time also in as a constant factor, but as we move forward, we'll provide the. So let's say people can just go and select like, which kind of a version of the architecture on the lower level kernel side table, they are looking for and they control that
00:50:00
Greg Osuri: so, one of the points I want to note is you know, I mean like, The balance between non-custodial experience versus user first experience, right? A challenge. We're seeing especially as we're, you know, venturing out into non crypto use cases especially and whatnot. There's a huge misalignment between what the users want and what the provided product provides just in terms of custodial versus non custodial. And, you know, I feel like the Akash Network core protocol and the core applications, and core interactions should be completely non-custodial. Whereas different clients building on top of Akash should be hyper focused on the user, so, whatever, vertical you're focusing on. You should be optimized for that need. So that vertical, right? That's said
Greg Osuri: When we are going into Akash AI, for example, right? Or like ML workloads, say I, you know, yesterday did a demonstration about like, Hey look, this is going to have stable, diffusion API access, most cost, optimized API access. I don't and like when you go to a AI then be like Go get a Kepler and get a bunch of tokens from like track and they be like, what the f***? Right. So don't write that notes, but use this. It's just one of those like massive misalignment and we've had that kind of misalignment between what the user needs because of all the product can offer, we're going to not succeed. So I would highly encourage folks building different clients to be hyper focused on what the user want.
Greg Osuri: And instead of like trying to push through, you know, a non-custodial experience, which may not be optimal, right? So that's where we draw the line. The core product call should be fully open source, and there's app, and fully non-custodial in any clients. We build on top of the core protocol. That is native clients. She went on custodial by design and fully open source, but anything beyond other clients that actually are verticalized and go to the users non-custodial. Nature is off. You know, optional, and I'll be very, very impressed if there's a fully non-costality experience that can be. That is amazingly well thought out in terms of, you know, In terms of matching the user's needs, I mean that'll be a dream come true. But I feel like we shouldn't We shouldn't let non-custodial, first sort of like be a hindrance or like a requirement.
Greg Osuri: Or a priority over what the customer needs. Just want to put that out.
Tyler Wright: Absolutely. I know we're at time again, I'm going to PR these notes so people can take a look at them. And then after they're committee, we can update go ahead and Neil.
Anil Murty: Yeah, just a really quick comment before we end then, I don't mean to put the Plato team on the, on the spotlight here. But quick question to you, Jigar, or you can think about this as well? Do you? Since your thing is technically a clients, it's not a deployment clients but it's a client for spinning a providers. I think it makes sense to have you guys, you know, be part of the sick as well, but just want to put it out there for anybody else to find about.
Jigar Patel: Yeah, so basically we started like that's what like we started attending this. See of any requirement comes in for us for like for our provider build thing, right? That we can. Give our QR response, you know?
Anil Murty: Awesome. Yeah, perfect. So, you know, we'll update the page and include you guys in there as well.
Tyler Wright: Absolutely. Again, I know we didn't have much time left so that cloud, most, and pay toward talk about some of the updates, I know, unless you have an objections, I would like to have them make updates if they can in the discord channel. Because I know we're at time and then we'll have them start the next meeting or if there's anything they want to talk about in between meetings and we can make some time to do that.
00:55:00
Anil Murty: Yeah, sounds good to me. And in fact, for the next meeting we're hoping to have one Join us as well. This is a company called Fleek, so hopefully they'll be able to join us next week for next time.
Tyler Wright: Awesome. All right, I'll have the up of the notes updated very soon. I'll put them in the proper channels and you all can comment and again, Max Jaguard, if you all have any or the rest of the paid our team, if you all have anything that you want to add about updates that you want to share with this group, please just do that in the discord channel, but I appreciate you all saying a little bit extra for this meeting.
Tyler Wright: All right. Y'all thank you very much. We'll talk very soon.
Tyler Wright: Goodbye.
Jigar Patel: I think.
Andrew Gnatyuk: Bye guys.
Meeting ended after 00:56:03 👋