Mina Protocol Mina Community Videos, 1 Dec 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CSCON1 - Mina Rust

Description

This is ChainSafe's talk "Mina Rust" from ChainSafe's 3-day virtual summit called CSCON[1] presented on December 1st, 2021.

ChainSafe is a global leader in blockchain protocol and infrastructure development for Web 3.0.
ChainSafe's website: https://chainsafe.io
Get into the code: https://github.com/ChainSafe
Join our Discord: https://discord.gg/CprJHjrWG6
Stay up-to-date on Twitter: https://twitter.com/ChainSafeth
Read through our Medium: https://medium.com/chainsafe-systems

A

um So next up is a fellow change safer that I'm super excited to hear from. um We have lerna jaborian who's coming to speak on our mina rust implementation.

A

um She is the project manager for our mina rust implementation and she spent several years working as a pm in the tech industry in toronto before uh joining us in the blockchain world um and coming to speak at chain safe, so uh super stoked to have her present um and I'd like to invite lerna up to the stage.

B

A

Hello, hello good to go the floor is yours,.

B

Let me just share my screen and get set up all right, so I guess I'll start uh so my name is lerna. I'm the project manager for the mina, rust implementation and I'm going to start out by maybe giving you guys sort of like what we're going to cover. So if we can yeah go to the next slide um a bit hard to see, but basically um I'm gonna give a quick intro of the team.

B

Then I'll talk a little bit about mina their history and give you guys some facts, uh then I'll get into the actual implementation that we're doing. Why we're doing this, and maybe why we're doing it in rust? We'll talk a little bit about the milestones and we'll also talk in detail about milestone, one which is serialization and milestone, two, which is block selection logic. So those are the ones we've been working on this year.

B

So I can give you guys a lot more insights on what's been going on, and then we can sort of end it with some questions and comments so yeah to introduce our wonderful team. We have elizabeth who is in charge of implementation, and next we have raul. Who is our team lead, there's willem who's, part of our amazing solutions, team here at chainsafe and two of our newer devs, who have been hitting the ground running high long and shin went next like this yeah.

B

So this is a maybe like a brief history for mina, so it was founded in 2017 by long-time friends, evan shapiro and isaac meckler, so they were called code on the beginning, as I mentioned earlier, it sounded too much like another blockchain. So to avoid confusion, they switched it and became known as mina protocol.

B

The idea these two friends had were looking at other blockchains. They realized as you're, adding more blocks.

B

Things are getting bigger and bigger and bigger and bigger, and maybe to a point where you'd come a day where you'd have trouble syncing with the blockchain, so they thought about maybe having a super lightweight super small blockchain and that's where this idea came from so now, mina is known to be the world's most lightweight blockchain at 22 kilobytes, which often gets compared to being the size of a few tweets, and it's also never going to get any bigger.

B

So if you're wondering how that's possible it's because they use something that we call zero knowledge, proofs or zk snarks, so what the zk snarks will do is basically, instead of having the whole history of the block. It'll take a snapshot which makes it super lightweight and small they're consensus. So again they touched on this earlier today, but it's called aurabor so mesico.

B

So it is based off of our boris genesis, which was the successor to aurovor sprouse, and this was created by the cardano team. So I believe for them it kind of checked. The right boxes would have the right decentralization properties, but because mina is sort of the first of its kind being such a succinct and lightweight blockchain. There was still a lot that they had to change, so they sort of made their own and then called it samasika, because it means small or succinct in sanskrit.

B

um A couple of the roles I want to mention is probably because it sounds different in other blockchains and you probably won't see something like a snark worker, so there's block producers, which are the equivalent of the validators and the snark worker is the person that provides that snapshot for you that you need and, of course, their source code was written in ocam, so this is sort of where we come in and why we're reimplementing everything in rust. So, yes, thank you for that slide. So I'll talk a little bit more about that.

B

So why are we doing this well? Having an alternative implementation for any network is good because it will increase resiliency and make it more robust and rus is a very developer, friendly language and probably one of the more popular languages in the space.

B

So it was definitely a great choice and having more people that can read and comprehend the code is great, because you know there's more chances of people contributing as well, so something else that we're working on this project is enabling in browser environments, so rust again, is a great choice because it can easily target web browser platforms. um It's also. It has great tooling, to support syncing mina nodes in a browser with the mena nodes on the network and the the last piece of work that we're doing is we're building out um non-validator nodes.

B

You can also call these nodes non-block producing nodes, so we're sort of laying down that foundation and groundwork for when they will at a point want to make block producing nodes um and yeah. As I mentioned, there are five milestones, and one and two are the ones we've been focusing on this year and we'll do the rest in 2022, so I'll just go through them quickly, so you guys can understand what they're about so milestone.

B

One is serialization so in this milestone, basically, we are using something called binprod, which is a serialization library, and this comes to us from jane street, which is a training firm, and what we're doing is we're basically allowing two nodes to communicate with each other. Next, there is milestone two milestone. Two is a block selection logic. So, basically, you are allowing the network to know which chain to choose when there's a long range or short range fork, um milestone three block verification.

B

So this is where that zk snarks would come in so you're sort of verifying the proofs in the blockchain.

B

We would be building pieces to verify how to validate blocks and how to I guess, verify transaction proofs on the mina blockchain milestone. Four um mastermind four is where we would sort of store and update the account database. So all the data will be stored securely, so we'd be building out the pieces to allow the mina nodes to persist, uh blockchain states and ledger information in a database.

B

And finally, we have the p2p support milestone five. So this is where we would allow the nodes discover peers by using the p2p.

B

So this would also support the submission of transactions via a gossip system um validate and support compatibility for those browser environments that I mentioned earlier, which would allow mina to be accessed easily from pretty much anywhere. So, let's jump to the next slide.

B

Thank you so much on one like I said um I'll start by maybe seeing what serialization and deserialization is so with computers um when we're talking about serialization, we're, taking or converting objects into bytes, which then can be stored or transmitted, and when we say deserialization we're doing the opposite, where we're reconstructing an object from bytes, so how we actually did this for our project. uh As I said, we use binprod which jane street developed.

B

So this is coming to us from the high frequency trading world. So you know that it was something that needed to be fast and robust, and what we did is we tried as much as possible to stick as closely as we can to the reference implementation and make a perfect recreation of that in rust. So we went through a lot of challenges because you know having such a complex type that has to be bite. Perfect is incredibly challenging, so our next step was to use something that we call cerde.

B

Saturday is just a rust crate that we used to implement serialization and deserialization formats, and once we had that we implemented a couple of traits that knew how to serialize and deserialize primitives, and then we added more traits that would help compose those into actual data structures like structs and enums from there we use what we called macros, so the macros you can use to annotate your own types and that will automatically follow like a systemic procedure to generate a serializer to generate a deserializer.

B

You have to first make sure that you have all the types already implemented and annotated with this macro. So what we were trying to do at the time was to implement the types to match the old camel implementation, and it had to be perfect again because if it's not, then your round-trip test would all fail. So it was like this massive challenge to have to re-implement everything in rust from ocamo, in a way that the serialization would be exactly the same.

B

We also ended up not yet. We also ended up writing our own custom survey data model and this allowed us to break down the serialized form of these blocks into smaller chunks and then test those smaller chunks individually.

B

So we slowly worked our way up the type graph which tim now you can move on to and um yeah. This is very complicated. So this is what the type graph basically looked like from amina block in ocamo, so yeah big challenge, but very rewarding as well.

B

Okay, we can move on so milestone two, so this, like, I said, is what mina uses to select the correct chain.

B

So there are long range and short range forks, so you have to be able to find a way to figure out which one would be the correct one. So we have three main pieces for this. The first one would be storing and buffering the blocks received from their peers. This helps in choosing the right chain by sort of seeing what's being broadcast on the network, and for this we implemented a library that we call the block tree.

B

So the block tree library will help in storing the blocks that are received from the nodes on the network. Next thing we did, and we are still doing, is implementing the checkpointing functionality. So checkpointing is basically how mina figures out whether a fork in the chain is a long range fork or a short range fork.

B

So it uses two checkpoints: they're called the start checkpoint and the lock checkpoint.

B

So if the fork falls, let's say within the range of the starter: to lock checkpoint um it'll be considered a short range fork, but anything before the lock checkpoint will consider a long range fork and in this case you'd have to use something that they call chain density to figure out which one is the correct chain.

B

So, um oh not yet so the next thing I will mention is chain density. So that's the last big piece of work that we're doing um so. What chain density is there's a few components to it, so there's something that we call the critical window. So when you have an epoch, there are different sections and the critical window is maintained by the chain and that's what helps you calculate the chain density.

B

So when we're talking about calculating chain density, what we're seeing is there are different slots in an epoch and there's something that we call a sliding window. So the sliding window will move from slot to slot and it will calculate the chain density at a given point in time and storing the minimum window density is what allows me not to figure out which chain to choose if it's a long range fork and the way to do that is the valid chain will always have a higher chain density than the malicious one.

B

So that's pretty much what we covered in milestone, two, I just uh you can go on to the next slide. I want to sort of end the presentation by saying thank you to the mina foundation and omon labs, and particularly the people that we work more closely with that are on this list. So something that comes up in our conversations is how easy to work with and uh collaborative everybody is so just want to say a big thanks to everybody and uh that's it for me. Jimmy.

A

Awesome stuff super great, uh thank you for the presentation and thank you tim for stepping up and sharing your slides. That was greatly appreciated, um and so we have a couple questions that I want to run by you. um The first from dante, is: uh were there other other languages that you considered like like, for instance, go? How did you land on rust.

B

um I'm not sure if the client considered other languages, but I would say looking at all the different ones. Rust is definitely like the clear winner because of some of the stuff that I pointed out. It's just a really popular language, if not the most popular one in the space, and it also has really great features for what we're trying to build with this project.

A

Sweet yeah rust is awesome. We love rust. I also love go-to, so we also like that. um So the next question would be do you believe mina will be lightweight enough to run a node on mobile devices.

B

Well, that's that's the idea actually, um so there are many different. I guess like phrases you'll see on their website, one of them is powered by participants and I believe one of them is like a blockchain that you can put in your pocket. um So yeah. That's the idea to have people uh access mina through a phone and it wouldn't even need to be a really high-end one. It could be just a regular phone, so yeah definitely.

A

That's that's sweet. I love that vita is I find mina particularly exciting, so I I love this. um Let's see. Okay, we have one more question that I see and it's um do you plan to have the node work, both in and outside of the browser.

B

B

Can we be more specific on that.

A

Maybe maybe they could type more? My guess is because if it's a representation, it'll compile that it can compile to wasm, so maybe thinking that they could be run in browser as well as uh like natively. Yes,.

B

I don't know, I don't know, maybe they have a bigger.

A

B

Quasim and arm are part of what we're trying to build as well so yeah. That would be part of that section that I was referring to when we're trying to build or enable in browser environments. So, yes,.

A

And the big question: do you have a formal project name yet.

B

No, we don't. We actually went through a lot of different ones, but I don't think we landed on one so yeah fellow chain safers I'll open it up, see you guys but nah. We don't have one yet, not one that we liked and also there are a lot of really amazing rust projects out there. So a lot of the ones that we wanted to use already existed so so far we do not have a formal name.

A

And also if anyone in the community has thoughts on what we should name it hit us up. Let us know we are open to suggestions. Well, lerna. It was awesome having you on here. Thank you. So much for presenting I mean is amazing. You guys do amazing work, um I'm stoked to see what the future holds. So thank you.

B