 So, I mean everyone knows this now, so it's more to give you a quick overview, especially planning and update, and I don't know if we can, and also like the, by the end, first there's some people, something that is this is possible. Please let us know. And then the next thing is trying to plan next steps, things that were, are worth easier and think about the production at the end. And I will, and I mean I had a few questions with you, and discuss but like, if you have any others, please let us know. I think that Michael already motivated why we are starting with these kind of sensors, we want to try and prioritize, so not to rely on a single from the target, but be able to have an effort and according to the set use case to be able to, to choose so that applications can choose so that the system, the best and be able to leverage the security of the whole system, so that just have an excellent system. So in the end, what we are building is a framework. We are building the pipes in order to allow fast and consistent subnet and leverage what already exists in what we call the real network that our target is the pipeline net, but if this is generally enough, the process can implement this. It would be very good that we also take the segmentation to start like parallel, I mean checkpoint to parallel and communication to parallel. And briefly and high level, this is how paragraph and sensors look like. The idea is that it's a framework for on-demand horizontal scaling, which means that when users or applications want to scale from a root network, they should be able to or we should enable a way for them to spawn a new subnet within some consensus algorithm in some state so that transactions are validated in parallel with the ones in the root network. So in bit and then we can have like what we call the parallel chase, like checkpoint chase, like what Sarah is doing, we can have a deep point there so that the root network also has a deep security with something else. But in the end, everything will be ordered with respect to everything. Yes, that's in everything that depends on the state of another. So it's not necessarily. Not necessarily, I'm just running our central sections that affect the state with my subnet. There's no word, I don't care about what's happening. So it will be kind of a causal consistency system. Yes. There's no consistency among subnets unless you do something. Yes. Yeah, but then when each of them check point to something that is ordered, you'll be established to go to order on these subnets of transactions. Only for this thing. So you interleaved the histories. Let's say that you are running a consensus here in one subnet and the others of that they're running in parallel, the states are completely different, but then you interleaved pieces of the state when they turn with each other because like you go to the top of the hierarchy, you order on some like place where you both. So that's just the pieces of the state. Yes. There is no need. There's no practicality in aligning the subnets. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. Yes. R trip. Well, in terms of diferença, so it's, how are connections actually what you check points? You can talk to all the bodies. What's really role could you do. I'm not touching the same things.'re nothing to the same. They're not really decided which is where you can because. Yes. So which is the definition of don't know. Yes. But it's a bit expensive. It does not matter. Yeah. There are things that we do independently as we put it. You are there in events, yes, so events that are to meet you or that part of that we don't care about. And there is like that we have a routine and any user would be able to like, for example, we start spawning new subjects below and also from child sadness you should be able to still keep going deeper. One of the key requirements that we force in subnet is this firewall requirement that you can see some sidechain papers. And in the end, the firewall requirement is that we limit the impact that an attack in the subnet can have over the parent. And this limit right now is on the total supply that has been injected in the subnet, which means that we don't enforce any guarantee of security in the subnet. We allow some users that are malicious or that want to force their own consensus to be able to run attacks in the subnet, but the attack that the subnet can have over the over the entire future by the security supply. This is a case for now that we only have funds exchange like type of exchanges one we have more attention we may have to revisit this property, but right now like there's no way of because I will show you the parent chain extract all of the funds that have gone inside and outside the subnet because it's a gateway to checkpoints and like to inject or take funds out, which means that once you realize that something is actually happening and that's more simply than the one that should go out is sending messages out, we just like cut that subnet so that's the firewall requirement. And the way in which we communicate between all of these subnets is what we call respect messages. And again according to the message goes down or up. We have like two types of messages the top down and bottom up, the bottom up they are propagated through checkpoints, because like parents are not enforced to see all of its cells but Charles have to listen to those parents that it needs to know what is happening over there and it's the two banks on the specific address and that's why the top down messages are straightforward. It's important to say that we assume that we still assume that all minors on a parent subnet are not minors on a child subnet are also minors. Yes, but not necessarily in the sense that if I want to spawn from so I need to see me depart a search, which means that if I want to spawn from the picture, it means that I need to be a minor or a validator in the, but if I want to start, actually, I need to go Yeah, that's still the case because if I want to spawn from a child, I still need to have to be able to employ an answer to that. Yes, it seems to me the assumption that. So, you're running on the five point way. As you go down, you see that you're running for your trust the node that you trust that run nodes of each line. I mean, what is it, what does it need to be a minor or a validator in terms of an arbitrary consensus for us. You, I mean, you don't need to be in the right, as long as you don't want to scale from the back, because I can see that so it will not like this. So, if I want to scale horizontally the route, I need to necessarily be thinking, which means that that I need to be a full node, I mean, I could be a light client, but I won't trust that there's some kind of stress. But if I want to, or simply scale from the top, I don't need to be in the. How do you listen to the child or top down. So when there is a checkpoint, because you need to listen to some. Yes, yes. I mean, it depends on if you can be a light client, the others, because the only thing that you need to listen. As long as you trust the, if you don't trust the parent, you need to be there in the tech, if you trust the parent, you don't need to be there. This is what I mean, there is a choice. Yes, yes. You need to do it yourself or you trust the world. Yes. But like child stuff always, I mean, all tells me to sync with the path that's. Yeah, that's for sure. And then, if you want to scale the route, you need to be a minor to see the other part, so that we don't need to power. You need to be a minor or a relative. Except we scale the expected consensus from father and we, so if we have a subnet that will have mine or say, actual minors and the ones that keep that consensus, you need to be a minor in both. Because that way you don't need. But you remember like we started thinking that we were only going to use five minutes for the subnet so we didn't want to use the power in the. So that's like change a bit because now we support any kind of consensus which means that if you are in the storage market, you need to be a minor in the team but if we are running any other BFT and tells are just other BFTs it doesn't matter as long as you trust that you need some information. Yes, I mean, that's why you sing with the plan for sure. And you can be like, I mean, in theory, you don't have to present with the BFTs there, you don't need to trust that you need to read from like. No, but like, yes, okay. I mean, we, for part for sure and for the mood also, but anyway, yes. Just so I know I had to think about it. So the subject is basically some SMR system where the output of it is just a sequence of some transactions that I either either. It's a, it's a complete AC dependence that so it's like you were running a section. So actually, you look like low level is this, if you have this subnet with some state trees, if you have a VM has to be the same implementation, it's the same but like, it's, it's all like, it's on stage, it's on content, so it's like. What do I have to care about when I'm when I'm a child. Do you think I have to listen to the power. So what I listen to is basically the output of the prime business sequence of orders. And like events. Just events on a setup. So, okay, the logic of all this framework is implemented into an actor and so you have like the subnet coordinate director, which is the one that enforces all of this logic of the framework. And then, whenever you want to deploy a new subject, we deploy an actor that depends on what we call the subject. That has all of, I mean, a set of functions where you define the logic that you're going to force on that subject, which means like, this is a second. And then this is the requirements to join with maybe as Marco mentioned, we want to start in a data center, so you're going to force that the delay is just that way. So you have these actors and all the operations. So then you use one of these subnets that has a sort of subnet coordinate director, which is a good way to interact with the rest of the system. And here, like at this point, the only thing that you have is a side chain. And then another two different networks. And in order to start interacting with the system, we need to be. Right now, we have to figure out behind it. But you put some collateral. And from there, you can start interacting with the rest of the network. But you only listen to that in the in the current to be in these two others. So you trust the ordering and like the stack in the in the current, and you only listen to events in that. And we'll see now that when you want to interact with all the hierarchy, many of the operations are not sent through your something but with the fact. Like, because you send them directly to the subnet computer or to your subnet. Which means that you need to have like these two legs in both networks in order to operate. Okay. I hope we will make more sense in a moment. But, um, Yeah, just put things into perspective. So we have right now we're listening to PowerPoint, we have the only one story market. Yes, this year you feel a lot of VM potentially, they will be in my thinking spin up our own store market. We can define the rules in the VM often smart contact to find out around quickly. And in this sense, we would be like our own group only listening to what is happening upstairs, or we would need to run the entire cycle just to check when it is like, what is the current phase. So it should be just imagine the world is like, it's scaling in terms of having only one, and then you have like a long service that is run on a separate storage market. Right, you can have it in the five minute, but if you want to scale in the number of messages plus. Yes, when I wanted to find where you go, you go to the subnet, you have the subject and you can even have the sector you're keeping still in the order how we're actually in the mood and have the subject when you start to market like, you know how many deal I don't know how you can remember the power after it's going to have the power of steel in the main net and be just check for the natural but like having but I'm not messages to update of the sector to keep in order so that these can operate the kind of we have some one, but still keep the consistency. If you see a use case, you know, you see a use case from, from your real life experience, let's write this down. Like, we can have one subnet for minors one subnet for payment channels, so that we are like test how a minor operator, but then it's really hard because the purpose of that so the couple, you can have the power and the minor actor somewhere and what do you do the deals. So, but that's the kind of thing that we want to do. Yes. If I have a storage market that I want to make myself with Logan wherever if I won't have to put if I am able to just spawn my own subnet and still like be consistent with what is happening in the, in the name. Okay, I'm going to close with this. So, right now we know there is a match for voting because we have all the new stores on where we originally do all this, we get here. Right now, somebody kicked that off, it got five million fill from like the initial question so imagine if we did it, we can do like 20 million whenever, and then you will need to manage that in our way. And this will kind of work as a potential to do so, but then what you need to figure out is the message messaging back to the public. But that's the thing, because like it's the onboarding depends on people onboarding or public so it's that the actual sector keeping still in the maintenance. Yes. I mean it does not like the meeting is going to happen at the top chain, which wouldn't, at least when we were talking with Nicola about this how we can become all this stuff. Yeah, but it's something we should discuss. Okay. So what we're going to do is a role model, which is censored. Yes, there is a need to transfer funds there, I guess, and that thing is doing the, yes, everything's right there, but here we can like the FEM who can do it the best. The proper way to do it. Yeah. And I promise this is the very last. You're talking about operations, transactions, messages and stuff like that. Is it all the same? Yes, sorry, because I didn't come down messages. Messages. Yeah, I'm sorry. Messages equals transactions. It doesn't equal events. Is it events as well? Events are different. Events is state changes. So you set the message which changes the state in an actor because it's my contract with the sector. And events is like a state change in the state. Which is, which always corresponds to the message. No, the message to your assistant. Okay. Yeah. And then like, I should use the state tree, which is like all over. So we have the chain. I'm going to state tree and the state tree is in the system. So we, you don't need the chain in order to have like a latest. And a message is triggered changes in the state tree, but you also keep the message. The state tree is very well set for terminology. We should rename it something else just for people. The state tree, I mean, the state tree is the one that I know I know less. It's, it's, it's, it's a typical, you know, which causes lack of power. Okay, so events changes in that contract. So yeah, like, if we want to deploy a new supplement from my child, the same, like we deploy a subject that defines the different policies and we should be able to scale. And yes, this is so maybe you have realized this, but we use deterministic naming for subjects at this point because that way you can work. You know where they can be deterministic in store. And also you can, and you see why this is interesting. And also there was another thing that I forgot. And I don't remember, but the thing is that the way we should name that is that the main net or the root chain is called root and from there, we, we infer the idea from the actor the idea of the actor because this is persistent so the idea of the actor that increments the subject interface for that. So the one that governs the subject in the past. And the reason for this is that whenever we want to communicate with another so in the transfer layer without we share the layer but each subject has some of those sub topics. So that if I want to so like the messages for our, we don't have to discern the different messages because they're running. And if you want to communicate with one of these subjects, we can discover them right away because it's just knowing their location hierarchy and sending a message to them. So this comes really handy when we need to access messages that have been started by using on the right. Someone sent me a checkpoint they don't send me a check point for all the messages I need to be applied process. The CID. So I have to have a contact resolution problem or to be able to fetch those messages from the right side. So we need this way of being able to communicate with other subjects that are potentially sending messages. What you will use here is to be around searching for certain you will use the like. We have what you're going to make the F for address that adds some information. So, I know that that's only for messages and see if it's just a fancy. So you asked for the fancy, but you know what, who to ask, what does it say to us because you have like this step for. But the F for address is any type of address with context. And I did like the, the way which we interact with two different checkpoints as I said is either like, or. I don't think this is needed. So that you can start interacting with the rest of the character, commit check child checkpoint, which is to commit because where you commit the checkpoints is in the back so this is the way we make new checkpoints from a child send messages because we don't. So, this subordinate is our statement to the character framework. So sending a cross message means sending a message to be separate and he said that those are all of the tailoring of the, of the message somewhere. So this applied first messages is called by the VM in order to apply first messages because then there may be some additional logic required actually there's some additional way required for last mile ordering. And this is done by my section because it has track like a single supply and so on. So the VM also runs a pre like check with from the SCA so this is called by the system. So this is something that we are, that is to the words, and it would be great to store state in a sector so that when we keep a subnet or someone that wants to create a snapshot can take the study of potentially the, I mean the state in some subjects it may not be that large and it would be a sector for sure. And it would be very good when we had a VM without like this native capability of storing state. We'll figure out how we can retrieve it afterwards but like the idea is to once a subnet is killed to have a way of having the state tree there so that we can have the checkpoints will still be on top of the hierarchy so that we have ways to verify and access this thing that has been killed. Yeah, in the subnet that this, I mean we have a reference implementation of the submit that the idea is that this is used to define. So, as long as you don't have these constructors where you said the jazzy, the subcontractors of how often you want to checkpoints will function and so on. But like the join the submit checkpoint that you leave and all of these functions are used to find so that any application can choose their own policy. And with the submit checkpoint here so so how it works the checkpointing is that, in the first instance, the size of the PKG by now in our reference competition we just use a majority so a majority of the miners, two thirds of the miners have to sign the checkpoint in order to be propagated to the SCA. But the idea is that anyone should be able to build their own because the checkpoint where they choose to propagate proof of state. So you take the state tree you do whatever you want to smash it on top so that anyone can verify they have access to the state or if you use say a computation maybe you don't even need access to the state. But. So the subcontractor contains this slide, what's the policy for. And so that way, like, we pre agreed upon the proof that we're going to use some of the signature policy, and then that keeps running and right now the reference implementation uses two thirds, but we could use a J, whatever, something like that, or checks this before propagating it and committing it into the practice. So that will be full-fledged for the proof that you want to use. Yes, and this is the part where we went further because maybe I don't know like we need to check, but maybe we don't even because right now our truth is super nice, which is the tip set for the effort in which you're just one. If you have access to the state you can figure out if it's correct or not, but maybe we can do something where you don't even need access to the state. Just by inspecting the proof you can verify. Of course, if you want to do stuff with the state, you will have to access the previous thing, but just for verification we want to make it not needed to access the state. And yeah, so this is, again, working progress, so if you see something speak up and also like there are some parts of the system that we want to explore later, so we have the price, and hopefully we will be able to come up with a catalog so that users can, for instance, like the consensus right now, it would be really, we can have a taste, a taste and so on, we will have a catalog to offer any stuff that gives ideas that users should be, and the aspiration is even that in FVM you should be able to implement the consensus so that you can plug your own consensus. So that's like the aspiration, but right now we have reference implementation of where we come up with this. I don't think it's really needed. Yeah, sure. Thank you. That's a question because. So right now, you have an interface to the messaging. New PFD is just a framework. I already framework that needs to be running. Yeah, but isn't that we know. Yes. Right. So if you have access to messaging layer, and it states, you could hold the checks. So you just need an interface to interact with the state tree and the messaging there, and then you implement the logic as part of the, of the BN. There's a terrible one. I should check. I don't think there is. It would be great. And it's like, you could take a new BFD from FVM and take it to not Solana but Polkadot and be able to. Just not. Maybe this is a great idea from. I was thinking about this the other day. I think you might want to wrap it up. Like for you, we have a catalogue, hopefully we'll give a catalogue for anyone to be able to have consensus that there's just there. And right now, like, this catalog is already there. But we've like nine super. People work at a non consensus. And we did it. I mean, spawning a subnet is just sending a tree like high level where the messages are close to the party, you sense, either the structure, or the joint you want to join our already in the summit. And when you are over the pressure, you register in the corrective and like the summit actor is not so the only one that can send messages to the related to the, I mean, like register, joining and so on. And it's not usually the one that is at the summit. And it has to be the subject actor with the right ID, if not like the message is not. And this is the way to be forward with information to a flat role, if I will kill all the same on the summit. So when you join, you join locally at the moment. So soon that there is a minimum number of something that may define a minimum number of people that's not satisfied you're joining your way. Yes, yes, when you do now you join, and you can keep mining in the, in the section if that's enabled, but you won't be able to register in the current. So it means that if in the summit, I'm going to say we need five minutes, you will be receiving collateral and you want to release that collateral to the SCA and register to the party on to the part of the start. Wouldn't I be locked with nobody else? Yeah, but you're locked in the section, you just do the section. Yeah, I'm just thinking about how to present this to somebody who's actually joined. I want to be there but I didn't wait for that. I mean, if you have to wait for, I mean, you can always like release your platform. And so you can't be confined from the U.S. We want to know. Yes, yes, kind of, you know, changes the way that you want to also. We can start with some trust in yes, yes, and that's why we so that is why this user defined. We have an example even in the presentation where you wait for three notes. So you cannot. Yes, yes. So we're going to right now the summit that is deployed and it's just one note because we don't care, but the idea is to have a reference also that when you saw the reference commendation that you choose a BFT, you have to wait for three notes in order to start like for the summit to start. What does it mean? Three notes. What does it mean three notes? There are two different minors that happen. That's what we mean by max. So minor need to have flat one they are at least, and there's a minimum amount of collateral. That's up to you, you should find so so to register there's a minimum collateral, but that's why this user class so that according to your sense. So basically we are taking some of the things from there. And it's like with this fireman it's about imagine how well it seems for some, you know, finding state basically you found the company initial shares and this is what you're like. And people can see that. And can this can make a decision where they want to try to understand what it is. I also remember. In which case the number, the physical number of notes doesn't really matter. We just like, I mean, you could look at us, but we only try minus because they have a lot of resources. Yeah, but is it at least three notes? No, you three different countries. Yeah. I don't know. Why would you consider me that way? I can't say. Yes, yes, yes. That's why you. It's up to you. So we even have a down so that's, you know, the supply and the submit. And that would be like the LSE, like, I don't know, you can't give you a bit of a sense of what is the size or what if it shows that the classroom has in the classroom. So, you know, like the supply, there's a million. Yeah, yeah. That's kind of a layer to approach where you see how much it's not good. We don't know how we're going to do it. Is there some definition of this? No. We have ideas where there's no. We can figure this out. We can figure this out. If you want to. We find the questions. We can have a look at that. And then. At least there are lessons if we propose something for them. Understanding what we want to do and helping you. Do you. Can you reuse your collateral? Like. No, it's not. Yes, it's not. So when you first came, let's think of. Okay. Okay. It's not trying to stop notifications every three seconds. Yes. You have the route. You want to find me. This means that you already put collateral here. That's frozen. You need to be a new collateral index here. And then. Okay. Yes. Just one second. I promise. I don't promise that the warnings are going to go away. I hope so. Okay. So. Really quick. Checkpoints. They're used for two reasons. First committee groups, like. I don't know everything. I'm from. So we hear you have like a basic data structure of. Checkpoint right now you see that. Really nice. The proof is just the airport. Checkpoint. And the. And the. The. The actual tip set. So that we can verify the. I've tried. It's quite. They try again. They don't give up. It's just like it. I said, keep trying to connect. Yes. Sorry. Sorry. Sorry. Sorry. Yes. The proof is this. I don't know what you mean. And then we have like information. I mean, I don't know what it means to be cross-messages. But this is just a simple question. I'm not sure what it means to be cross-messages and tell checkpoints. I even to be cross message that we don't probably. Of the actual message because they could be huge. What we probably. What we call the. Method data. If you've seen the lots of these products. This message. This is the message. But. We aggregate all the CD's of the messages. So that you have a way of watching them. If we need. And see. Yeah, like. are free to choose their validation logic. So how does the checkpointing work? When you define the checkpointing period, so I have 100 epochs, and every 100 epochs we want to make a checkpoint, but we always to be from windows like versus the checkpointing window, which is the one where the checkpoint is being populated. So all of the trash messages that are that needs to be propagated and all of the child checkpoints that are committed are populated inside a checkpoint template. And then when we reach the network for the next checkpoint, what we open is the checkpointing period for the next one. So we start out with a child checkpoints and trash messages and the sign in window for the previous checkpoint, which means that here is where miners will get the template at the root of the speech miner is supposed to approve in the current implementation, sign it and commit it so that the submitter can check the two thirds and what the two thirds are checked, the subject, sorry, the checkpoint is committed in the current. I like that. So how does this bring, who sends the transaction to the third? No, every one. Each miner. I mean in our two thirds, that each miner needs to sign one message to the first. Okay, but I hear 100 nodes in the period and 10 in the... Yeah, but only the ones in the second. The last one or whatever to send out. Only the ones in the second. Yeah, you need to send the first one out. Yes, yes. So that's a lot of message. Right now they are. That's why we need to figure out another one. You won't do it right now, you will do it right away. You're going to send four, I'm asking for more and eventually it can reach this deterministic bound, but hopefully you're going to... Actually, we're on set, but everyone picks a coin and... Yeah, I will do it right, okay. And I'm going to have to keep the coin with one second. And even more, like even if all of them are malicious nodes, we can... They're just fine. No, wait, wait, they can't because like even if you're sending messages, the actual checks, it already has been completely required and it doesn't accept new messages. So it means that's it. I mean, if we reach the two thirds and you're still like, I mean, the messages, they want to get in. I'm getting in, but there will be some work. No, no, no, because you... I mean, we would have to check, but like right now as... This is the worst case, you got all the malicious nodes. Yeah, but like it won't even be proposed as message because before proposing messages, I thought how in fact right now you need to execute them successfully. Which means that I would try to... Of course, I could push one, but no one would accept that one. So it wouldn't be even executed or the messages would be involved. So because of how it's architected in the pipeline, this is not the case. You only would have like two thirds, which is already two, like a lot, but once you reach the two thirds, you wouldn't keep accepting new ones. But we can see these low level cases. Because we've literally mentioned this, like you... So you can coin, you use d-lan, and you choose like... Good, if you keep it a little bit coin, that helps in the honesty to reduce the low levels, markers and parameters. If you use an actual random message, you can call the VRF, then you can also control the few messages. Yes, you can even equal the ticket from d-lan. Yes, and then... And if you allow it to the actor, you will... Also, there might not be enough, and then if the malicious, there are malicious ones, it's not about the operator, then it's okay. You don't need that much, I repeat. Yeah. Think of a sequence around there. Yeah, yeah, I see. There's another thing we have, like, the sign in window is... The checkpoint is equal to the checkpoint view, because here you have 100 levels. I mean, that's why this is usually the point, because according to your use case, you can choose, like, maybe two places, okay, 100 levels, and it's up to you how you, like, yeah. Right, here's the checkpoint view. The best way to see it is, like, the signs aren't as, like, as climes on the bottom. Yeah. There's nothing different, like, if I'm a client and I need to say, uh, yes, and I don't want it to be censored, I have to do something. That's how I always thought it was. Yeah. No, no, yeah. Yeah, because you always say it's possible. Yeah, yeah. I'm sorry, but with climes, there are not a lot of climes. No, that's how I thought it was. Yeah. Yeah, it is. And, uh, yeah, so if we don't get the message that I'm saying, I mean, Schrodinger probably is getting it in place in order to... Detectors, everybody gets five minutes. To go through it, and then, like, once you sign it, the signing and the submitting is in the past, so you submit here and one for two thirds in our reference or whatever. You're recommended. Uh, it's fulfilled. You propagate to the... to commit in the... in advance. And leaving a subnet is like the same. You send the submit actor the same way that you join, and these triggers are released in the collateral with the caveat. That is that if you release collateral below... So if you're releasing collateral, the collateral goes below the threshold. You... the subnet gets in a processing in which you cannot interact with the sub... with the rest of the hierarchy until you commit to the collateral. Either joining or... or adding new collateral by the existing matters. And finally, the cross-net message. So I think we have two main cross-net messages. The top-down, which is like from a parent or from the root to a child. And you have the bottom-up, which is from a child or someone of a download hierarchy to either the top... No, to the top. So the first ones are just, uh, committed through syncing. And the other ones are propagated through the checkpoint. And then when you have a checkpoint committed to the parent, so you can, like, run it. And then we have the... what we call the platform section, which is just a combination of top-down and bottom-up, which means that if, for instance, you want to send from one channel to the other, you first will bottom up. So you propagate the checkpoint up. The SCA here detects that it is not directed to this, uh, something that it needs to be directed. It sees that it's the top-down the one that needs to be directed. So it will, like, prepare a top-down message to send down. And here, the same cross-messages are sent to the SCA. So it will actually send us a bottom-up. It will go to the SCA in the subnet. And it will checkpoint in the top-down, then commit it. And then the top-down is, like, the same way, the same cross from which the subnet from which you want to send it as a message. And the actual commitment is done. So we have two message pools now. We have the message pool for the subnet, which is, it's a message pool. And then what we call a cross-message pool, that is responsible for, uh, identity, like, please send it to your vets, saying, hey, there's a new cross-message that needs to be applied to this subnet. And also it means solving, because some of the checkpoints on cross-messages are going to be checkpoints. You only see the CID. So these cross-messages pool is responsible for the resolution that's the CID. So that you have applying messages that you can apply to locally in the messages. So for the top-down messages, like, you have a bunch of messages that come from the parent or, like, the root or whatever, that you, when these messages trigger the state change, and this is the event that I was mentioning, the cross-message pool from the tab will see that there are messages with new announcers that haven't been applied. So it will propose them as applying message in the blog. It goes through the contenders, the blog is committed, and then when it needs to be executed, you have a way of identifying that it's a cross-message. So you apply message in the SCA before, like, executing the message, and this trigger the state change in the, in the corresponding set. And from bottom up, it's kind of the same with the difference that you may be propagating not only cross-messages, but also, like, cross-messages that come from the childs because they have been committed and you need to round them up. And all of this included in this CID meta, which is, like, the CID of the list of things that need to be applied somewhere else. You propagate them up with a to and a from. So, actually, what we're building is we're calling the meta tree, which is actually on the front too, and the list of messages that need to be applied. So you look back at your piece, you take the CID, you propagate it, and then this is a recursively resolved by the cross-message pool when it receives one for itself. We will propagate the message with the cross-messages. The cross-message pool, in this case, from the parent, because it's a banana, it will see that there's a new chip that hasn't been committed with cross-message spending with messages that haven't been applied. You only have the CID. This actually is not, messages not applied by the CID meta that hasn't been applied. So you get the, it will go to the subnet, I mean, call the content resolution protocol and get the list of messages behind that CID. And the rest is the same. We include them, we propose them as people's other messages in the processes, we commit it, and then we apply it. This is the CID applies it as people's any other message. And the cross-message resolution protocol, it is, so it's quite straightforward, but we have a push and a pull approach. So whenever we sign the checkpoints, we also push here to the subnet where we're sending, eventually we'll get this information so that they can cache it locally if they want, if they don't like this to them. So we send the checkpoint and we push that information to the corresponding subnets that will need it so that they can cache locally if they want. This is probably what you're gonna use for two points and this is the protocol that I was suggesting you to use for each to access the storage. So do you guys get this part? Because you properly, to checkpoints, you properly the hash information. Where is the information? It's in the original site. There are two approaches. When the destination subnet gets from the parent down the information that something is live on, it can pull the information or it can push from the source information today. So, and we're leaving. I would talk about bailout of the transaction. Bailout of the transaction. So there is, basically you get the information and it's addressed so from this subnet, you get as a destination from the parent something is going on and this is the hash of information that actually gets sent in. Where is the information? And this is what we're going to use today. So when we push it early, this is in the, I have a central level which means that someone mentioned not to even listen to that message when we started. So it's scalable in that sense. We have even by agent to avoid flooding. And then how that is that at one point, like we need this information because we have a cache. We are like seeking from scratch, whatever we have a full approach where we send the message and we say, hey, someone here, please send me this information. And we have a bunch of like flooding, including filters and validator filter and so on to avoid spying, to avoid like the only thing that probably is missing in this protocol is that one to one like a huge fear because as we have been thinking right now in fact, because if this fails, you don't have a way of reaching. So I mean, if you've got some sort of fail for some reason or others, we need to have one to one. The problem is how you discover someone in the right subject. That's why we're using that stuff. But like that would be the fallback missing for BitShop, maybe, to see. All what we built, at least what we're using was the ability to be in this network. So don't do your own method instead. And everything's going to trust. We use IPLi for all of these checkpoints and all of these things. We could call them Marigold Forest. Eventually what we do, we solve from other stuff next time. And yeah, this is something that I think we have a session so I'm not going to go deeper. I would love to talk with you. And have your input. And that's basically from here, like the things that I was thinking is first, what are the steps? I mean, all this is the method. Extended as an execution. Everything is implemented. And it could be in the form of water, so it's worth. Now we need to think we want to invest time on trying to integrate computer work data, trying to have extensive testing or integration of the experiments or start thinking about translating it to MPI. So what are you going to show is the cross-summit transaction. You want to show the data? Yeah, cross-summit transaction of 5.8. Of 5.8. Yes. So these are big, these are one-way things. I pay to you, but you pay in another subject and it involves upstairs downstairs. Resummit has to do with upstairs. It's done somehow. I don't know what that is. It's a movie. Well, even the address. Probably not this one. Okay. Yeah. But this is the address that they're all using. So the latency would be, you know, the latency of a cross-summit transaction would depend on the checkpoint latency as in the layering of the... But otherwise, it should be okay. I mean, we never need the first yet to perform this test. A 30-inch later upward to being away from the next checkpoint to properly. Yeah. Yes. Yes. Okay. Thanks. We should... But, but, but, but in principle, in principle, you could, because of this cross-summit push. If you're pushing messages or not on another... Then I'll talk about the type of test. Yeah. That might make a way. You could have a pending information that the payment is pending like really quickly. And then you're just like, you know, you get on your mobile app in the future, you get a hot somebody pay you, but, you know, it's pending for one minute more. Yeah. And then it's not... It's also similar to massively. Like your transactions processing pending, whatever. And then you wait and you can back down there. If it exists, you get something. Yeah, you can. That's exactly what you guys are asking for. I can take the checkpoint here in this case. Yes. Yeah. What? Yeah, exactly, because there's all the batches that you guys can't find in the distance. Yeah, yeah. It's a pretty smart protocol, but I think it's... But it works. I mean, there's no reason we should work with it. Okay, let me open it. I hope it's every base we're going to interact with. Oh. So we don't know if it exists in this computer, so... Okay, I need it. So we're really developing engineering to a level of engineering, not like very high level distributed systems concept. This is... As I see, each of these layers is basically an SMR system that has its own log operations at some state. And from this point of view, it's the application level that stated that the kind of new people are updating each SMR state in a way that makes sense. There is a structure if you want to... Is it we have a transaction from one SMR state to the other which would be a presumption? There is a certain way of doing it. Yeah, exactly. And this is specifying the way. Now, again, what we're currently going to see now is if you want to do atomic transactions process... Yes, that's not really good. Yeah. And there is a... And of course, if you progress on that, can we have a session on this with tool and understanding you know, if you are happy with it? When you say forklifts, it's a little... Yes. We need to describe it as a forklift, where partner-executive consensus is not just you know, take money for something else. What do you want to do? Now you're changing that model for what we're saying for whatever reason. Yeah. You're taking it for a cross transaction. Yes, sorry, for a cross... Cross-net. Cross-net. Cross-net. Cross-net. Cross-net. Cross-net. Atomic thing. It needs to commit at the first point that's access to the right. So, how do you do that? You can't say... Are you going to wait for... If you're going to increase the size of the technology... Well, if each of the two subnets... Yeah, I mean that... I would usually say the atomic thing, I would always count it in this... Five or the other, I would say. And then... And then... Yeah, you're sort of on the JN, like... Let's say, kind of like... Perth assignments, they're not going to go up. Yeah. Or you're going one way, one way, because that's a model. So, you're going to just be pushed in a real program, a forklift. Why are you doing that? That's it. You're going to have to change the model very quickly. It's consensus on that. Oh, my God. In conventional state, after all that happens, and then you can go up and change the state if you're going to access it. It's just used to facilitate the communication. Yeah. It's not a huge thing. The cross-net interaction can change the state of whatever it's actually going to be. It's just there as a common law. Yeah, but you were once up a time in a different summit. And we had a common ancestor. We had a common ancestor. And then we want to do, like... And there is what I want to mention. Well, each of us, that or those, you can't return state after after the summit. And I will send my state after after the summit. And then, would you need to tell the common ancestor, at least confirm that this is what you committed? Mm-hmm. We'll change the other party, of course. We will have the common ancestor. No, we check that they're the ones for each class. And only in that case, we tell the other person, okay, I've got a question. We're checking if she missed it here. Yes. Yes. And then we just let you know. Yeah. Yes. It's like these trusted third parties in this multi-party communication. Yeah, anyway, I think my point was we're going to do it from a sense that we just need to agree that we're so much more. Yes. You're going to be able to do this next. You're actually going to be able to do it. I'm going to, like, yes. Yeah, I'm actually going to do the same because the part only you do is by definition. You can both do this. Yeah. Yeah. Yeah. Yes. Yes. So we're going to, I might just start. Yes. Let's make that one big. And then we can ask you. We're going to start with this. I'm going to do this sometime. I'm going to show you that. Okay. I'm going to give it from you. Okay, so I'm going to show you. Okay. Here we have it would be called or to route it. Mine. And this is like what we call the delegated processors, which is just one, mine and mine, a long way to create something else. Or we're going to at least deploy a new subnet. A new subnet with, I mean, we have human-readable links because we have even a discovery system and we have human-readable links for F and S. So we were discussing and we're saying that we want to do it from the parent and we've been trying to do it from the one which is the subnet. So we're going to have a subnet and a subnet line to see what it looks like. So you see that we just deployed, at this point, we just deployed the subnet actor. And it would be slightly, so it says that we could have an environment subnet in this address, but it's not registered yet. So if we use list subnet, you see that there's nothing we want to register it to the parent. Now we're going to join it adding two quadrants of collateral. So we just take the subnet ID and we add two of collateral. The minimum I think is two, so if you register right away, we don't have it. So now we should see once the message gets through, we should see the list of subnets. We have now the subnet that is acting and with the same two quadrants, but now we circulate the two quadrants. What we're going to do now is mine and we're going to see that so here is some other identity. So now we started mining in the subnet and we see that we're going to stop here. The main chain is proof of origin. It's like slower, but you see that we just mined a plot where mining works even without mining in the root network because you're sending the subnet. So we have two different chains, two different indices, so they don't have one in here. When you say mining, you're mining on a subnet because you're committed. Mining in a subnet. So now we're mining in both points. Exactly, but you've committed to the two fields. Now the operation downstairs is happening on a subnet level. Yeah, yeah. So this is mine. You might notice exactly whatever subnet. So here I'm sending the subnet. Actually, we can do something. I mean, if we do well at least here, it goes to the root. You see that I have 174, but if I use subnet 8k. Yes, okay. Yeah, that's the first one. Okay, yeah. You see that here should pass you. Yes. You see that here I got 174. So we have these two chains. And now what we're going to do is we're going to do a top-down transaction. So here what we're saying is that we want to inject from the root see that I'm using the subnet 8k thing too. So from the root I want to inject in T0, 1, 0, 0, 2, 5. So it says like, business needs to be validated for shortly in the subnet, because this is a top-down. So it's committed on top. I want to commit on top. This is super fast, because like it's just one message. So we should see here if this is well at least. For the subnet, you see that I have a 2, 5 here. Yes, on the subnet. On the subnet. Here in this subnet. And if we see the security supply, so if you use the check point for the subnet, you see that the security supply now is 2, 5. We can also release. So there is, these are, instead of sending a complex purpose, it's fine, I mean, it's just one way up. So up and down. And now we're going to do a proper one that explodes through several subnets. So the main thing, the supplier, recomputing all the time, the latest state, or it's fresh. You can see it's like the moment you apply so in order to apply a cross message, the application is done through the SCA. You can see this message, let me mention, which is not done everywhere, it's just done locally. So in this implicit message, it checks if there's some, some funds moving and it's updates like state of the, SCA. SCA, yes, yes, yes. So that's why we apply messages. And if it's the text that's funds that are over the security suppliers, are trying to meet the subnet, it kills the message. So it doesn't have a message to report. So this is our final product. And we can show that. So that works. So yeah, we're just, we're going to release some funds. It's still going quite the way, so that we see another update. This is a bit slower in the sense that it has to become a regular check point. So if we list here the checkpoints, you see that in the next checkpoint, we should have cross message equals true, meaning that there's a cross message here. It's true or worse. So like I have 10 points between checking the verdict. I don't know. It's something. And you see that cross message is true, which means that it has cross messages. It is committed that and now it's going to be applied in the top. So we should see now that the security supply of the subnet is once the product was changed, it released some funds. No. Yeah, yeah. So now the security supply is applied right before the messages. I'm going to add really quickly a new subnet so that we can see how they wrote from one subnet instead of just up and down from one side to the other, so that we see the progress. So now we should have two subnets. Both have the word. You see that here, security supply zero, what we're going to do is send. We can see this list. Subnet is only made for the route change. Yes. Yes. Yes. It cannot be. It cannot be. It actually released subnets only lists because we don't have a list correct now. It only lists your chance. So if I go to the subnet in the current, so once we know it will show up. So we discovered it wouldn't work currently but you would expect me to give you it would. We are convinced that it's basically encompassing route. Yes, yes. But actually we could have a really simple comment because you can traverse the current. It's just looking to the state and say, hey, who's your child? Who's your child? Who's your child? So we could even graph the priority. Anyone could like grab the priority by asking. So I mean, that's why I'm going to comment that if someone wants it, we can have it. I didn't think it would be useful at this point. Someone who wants it. That's also a lot of questions though. Because you're just as pretty safe. Somewhere in the last thing you know you've got to send. No, it's like that. I could have like I, so I can see the state. Directly. Instead of thinking I'd say the state. It's one of, and it's protocol, like if it doesn't go through the, there's no state changes and running for me. I don't know, get an answer. I mean, I'm on a standoff. That's a good one. Yeah, yeah. I'm on a standoff. I'm on a standoff. I'm on a standoff. I'm on a standoff. I'm on a standoff. First comment. We cannot. Yeah. We cannot speak with this. Just about my version. That is actually just a little bit of a plow as a proof that you can see. Yeah. Okay. No, you're just asking. Yeah. So let's see. I mean, let me check what wallets are from here. And obviously I said, okay, let's send it to the same wallet. I'm going to send it to my same wallet in the, in the other stuff. So here I'm sending from the first one that I created. So what's the format of checkpoints currently? Does it assume something specific to file calling or? I mean, yeah, we're putting the tips in, but we're putting the Marshall tips as well. It's just a blob of light. We could put any of that. So to do this a bit more in the future. What do you mean? And because you could do that, you could do this form of submit and reserve the product of the transfer on specific address. Yes. But like we need to, yeah, yeah. I mean that it could be done and in many possible, but we need the way of doing this so we don't have yes. Yeah. Oh, that's it. Yeah, that's that. If material would be straightforward, but like in the kind of we have the same product, so it's how do we do this with the complex? So the question is, if I'm doing it off time, I'm going to be putting the key of the others that hold it for a while. How is this managed? So to say, whatever. But I don't think that's Yeah, I don't want to say Yeah, the first one is the practice you need to forget about do it in a specific way. We would make the DKG or something. I mean, if there's more something to be done, like right away, because there's not that much to see with 100 addresses, especially for like that much. So yeah, I'm going to send the 0.2 so we should see. Sorry, there's the question. Is the kick set, it's like the last message of the checkpoint it's the head. Okay. We don't have a lot. We have several heads in the chain. So it's all of the correct heads. Right. Yeah. We follow the message. So the set of the tips in these heads. Yeah. Did you forget set changes in the code? And then that is how we determine strictly order than when you need it. Yes, yes. So yeah, these messages that I showed you, it should have like we list the checkpoints. We see that this one included to look like the checkpoint up and then in this word it will be in the subnet in the subnet. One, we should see 0.2 without seeing it. So I think the other way around it fell because I was sending from the subnet that I know. No, wait. Yes, I think that we should see it through over here. Yeah. The last one. No, not this one. The previous one which was here. Yes, I actually that is now we can see one of the see so you see that here in the list of this which would have 0.6 in like shortly once the checkpoint is up and down. So we should see here. I mean, I should have chosen something that is not for the work. But hopefully after 10 months, we have the truth here and then we should see shortly that in this subnet we have 0.6. You don't see again. Now we're going to kill one of the subnets and you'll see that watching here. So here we are seeing the subnet in the right. Like sorry, the checkpoints. You'll see that once I kill it, there won't be checkpoints because I won't kill the swing practice. Well, what happens here? It's mostly push of transactions. So this is how they come. Or there is I mean, it's working with push because like everyone is listening. But if you want, I can start a new subnet and it will be able to see from stretch because it will be moving from the right side. So we can so we have but right now it's as we are pushing, we're just so aren't you so ready to do it in some sense? No, because like no one sends the message back. So I'm yeah, I didn't send it. So, so no one take please. So if there's someone who's on this that yes, if not, so now I'm going to leave the practical use here. You'll see that the subnet should be frozen and the checkpoints should disappear because yes, no more. I'm not listening to the subnet anymore. From the node and by this is active because I need to solve my state and they just have one minor response state. And the problem is that this is what we need the state because there's still security supply that can't go out because there's no miners for the subjects is what we mentioned the other day because which it would be good to have because we have a check and we know what happened. We can't but we don't have access to the state. It's minus. So right now it would be a lot more. It would be locked. It would be locked. But like I mean we can enforce that you can't leave like there are lots of solutions we haven't agreed on Yeah, perfect. The problem to the checkpoint is not actually anywhere. You know it's just a fact. So you've got to say I'm going to go ahead and solve them to that. Yes. I mean you have the check part community in the bank. So you have you don't have the data you don't have access to the code. Yeah. Yeah. You have a full check point. Oh yeah. You have access to the check point. Like between that there's nothing that's not there. No. Yeah. So just keep going on. So you don't have access to the data. And what I'll say and we can move on Marco so just like let's let's see from trash so that we pull all over the data I'm in the right way. And we're just going to sink so you can sink with any subject we're going to see with this actually let's see with the parent so that we need to pick all the check points from the chart. So now when I connect the two nodes we should see how it so in the top right you'll see how it starts sinking with everything like all of the check points it will be picking up all the information it's sinking with the check point with the routine and then I can sink with any of the channels and it could be used that's the see this actually if there's nothing that you mentioned I try to pull but if I can't find the information because I just killed the I killed the silence so we cannot sink because it could this is something that we should fix because I tried to sink with something but now they don't have us to sink with it because there are no lines so if you see something yeah I can pick it up with the content resolution but if you just come in with two events it's like ideal or something there are messages that we're directed to the top but I don't have a way of getting the check points so that's why it feels actually like join again that should work where's the sink when you do the research so you just cannot access sir is this is the same preserve because it's so big but you're not going to see it yeah because I have an it's my tears so I have a question I'm shaking I'm shaking it's a little bit yeah it's over the car but like not actually yeah no if someone if it wasn't me and someone else like a clutch one had to change sink we could the things that I was the only one and I disappeared so there's no point and now let's at least see the button is interesting because we have where to see because we have we should right we should have that functionality right I information so from mvm we can have an actor that's going to be yeah and we click on it so once mvm comes I'm taking this for granted even if we cannot reach it again what is that that's what Alvin is best at the guarantees that it's there it's there you can find it