All Core Devs - Consensus #23

Transcript

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music] [Music]

okay the stream is transferring over pulling up the chat box chat box from the internet please let us know when you can hear us and otherwise we're gonna get started this is issue 68 on TM repo the agenda I imagine everyone seen it it's very similar to it is how it is every week and we will get started with testing and releases so somehow before I went on vacation last week I messed up

the tests and BLS tests were not included subsequently they have been pushed into master but not cut as a new release we've also added some additional SS the generic tests I believe we're going to the intention and we can discuss this is to cut a zero rate three on my NER Tuesday release these new tests zero eight three is entirely non-substantive and includes additional tests and includes some sse clarifications and maybe like a typo fix and something else with the intention that this is the and it's substantive ly exactly the same as zero two but this is

the interrupts target especially for our September gathering it's anyone opposed to that this is that again it's essentially zero a to but it has some more tests great so that is coming Monday or Tuesday with as always detailed release notes and an associated bump in the test circuit cool proto is there anything else with respect to testing that we want to discuss for the 1/3 reduce or spinal feedback this moment this little change

or any small thing reading but those who like best and then I'll look at if we're really gonna do release yep anything last minute let us know soon we have commitments from every team and from the researchers already on this back then again continue only true yep and as for other things that are on the horizon for releases the deposit contract we run a verification is doing a buy to buy code verification deposit

contract they found a VIPRE compiler bug which I can share it a little bit that is to in the works of being fixed on VIPRE side and in addition we're going to be adding a checksum that essentially you locally pre compute what the sec root of your data should be and add that as an additional parameter these two things will be added to the repo sometime in the next probably two weeks but I don't think we're gonna cut any releases so it'll stay in dev until like mid to late September in mid to late September there are a number of small substances fixes substance to changes

that might that have come out kind of an audits and thinking that will probably come out after interrupts one of the most substantial is to remove the machinery that's currently in place for light clients as it doesn't quite do what we wanted to do and it adds substantial hashing overhead so this would be a deletion of like active index routes and something else so as a pure deletion shouldn't be too bad but I want to let you all know that is on the present as well but we won't be cutting that out until after our interrupts blocking any questions on that I will probably be removing transfers so

to say my substantial removals we can discuss there's generally some like minor issues that there it could probably be more elegant and we don't plan on enabling it at the initial fork so we're just gonna pull it out and think about it's more hey Danny out of curiosity are there any links to talk about that talk about the Putin upcoming removals I believe so and if somebody wants to dig around and find any conversation about that and Sharaf that'd be great because I'm gonna move on to the client updates and I can look around - yeah absolutely and I think I think there's an issue

around transfers that's disgusting like a more elegant mechanism to do it there might be another issue around removing it as for the other stuff I think there's an issue as well sure that's room oh okay so that's testing and releases next up is our client updates and today we're gonna use their beat nice did you see the there's a Moloch

RFP for alligator client independent of these clients so you could look into that I think that's just a GUI I read the RFP it's not just a GUI it it's very flexible and I think that there's also there's actually proposal right now but there's probably room for more it's very flexible it's pretty much do some of the things that clients probably aren't going to be focusing on so you can pull out a bill about your client you can add a GUI to an existing client you can build out monitoring tools there's you know there's a lot of value add yeah

cool thanks you have a harvest API that we have specified also we have connected our main executable to logs to be and now a chain storage is dropped on disk we have also worked on if one integration and implemented client that based on ERP see the JSON RPC credits to web 2 J however this PR is still waiting for get nourished also Anton has

finished a gossip sub implementation and as a part of GBM the p2p work and now testing that's implementation with with the demon from what I know ash hunt has finished noise implementation as well and looks like phase zero of a GBM implementation is close to to to the release all that we need is Java API and some infrastructure things to be done to get there so hopefully we will be able to use this

implementation for the internal process that we are having in September we have started to work on zero eight to upgrade of its back implementation it involves a lot of things to be done in terms of testing so it's work in progress and we have started to work on on the QA started for a little bit from probability testing and started work on Pfizer it's gonna be a become change father for the beginning but eventually it will work with the network stack our

implementation of the network stack as well yeah that seems to be all for from our side also thank you about Trinity High recently well mostly focusing on the requirements of the inter up according to the network clean snake and most work is in the pilot PTP module alex is working on the sick I Oh module for the in top usage and there are still some minor inconsistency in the pilot

appeal ah we are fixing right now for the wire protocol side we are also building a beacon receive a server we call that server handling the while protocol message request and that's an open peer and its will emerge as soon as possible and also the is covering five side that planning is working on the discovery five functionalities for the Trinity is one client but I think it would be great if I can be refactored to the east to later

yeah when it's finished maybe we will talk about it seems Felix is here today later yep so that's all thank you great thank you how about Artemis good morning all my name is Johan I'm gonna do a brief intro because I just joined team Artemis few weeks ago Thanks so I'm very excited to be working on e2 now previously if done database related blockchain stuff and sand castle which was a Scala theorem smart contract

language anyway so I switched off off of those things on to eat too and one of the first things I worked on was let me to be with Anton with the noise integration and now on to D Artemis updates Steve and Jen have been working on SSC debugging state routes as well as the serialization John has a pretty neat reflection based API or process for doing the serialization and deserialization Joseph is working on state tests and Johnny and plugged in a lighthouse the little bit into Artemis

so that we have a rustling p2p implementation of in Artemis now we've also gotten some benefits from Adrienne helping us Adrienne Sutton is working with Pantheon and he's given us some of his cycles to start production izing it with the metric side and that's pretty much it nice thank you what minor the last thing if Joseph is online if anybody has questions about the upcoming and drop please get in touch with Joseph DeLong Thanks

yeah cool and maybe in a little bit jezus can give us any relevant details anything any updates cool thank you we're going to lighthouse hey guys I'm sofa lighthouse we've got a few points that we've been working on over the last three weeks I guess firstly we've been trying to optimize our database I think we plan on going down like the the goth route where we have a cold optimized database and a hot one between finalisation ebooks we've been working on our test net stability in particular looking at extra metering and doing some bug squashing and because of the update

to the well the latest aspect that we merged last week we've been updating our networking stack and at the moment we're at the process of modifying and improving the syncing algorithm to match with the new RPC that we've got so we have to finish that before we're completely up to date with that f2 with the new earth to networking spec we've also been working on BLS standardization improving that and in connecting lighthouse to s1 with the s-1 deposit contract and finally we've been improving our HTTP API hopefully we're going to be transferring over tasty HTTP soon also between the validator client in the beacon oh cool so that's it from

us thanks guys great thank you glad to have you in a same time zone as well yeah it's refreshing it's not enough midnight okay prison yeah I can you just give me yep yeah so there's a few unparallel efforts going on so we're working on ramping our current sting and p2p service to align with the latest networking stack and we also respect her RDP for make you like very easier to filter out query for example I go on Korea blog industry you

can just filter the apparent load and above certain spot and then we also plug in the new for choice which is we refer to the for choice on a finalized spec and then working on benchmarking optimization so hopefully we can see a while on second slot thanks to and I will also implement the various obviously matter it's people are scared stations and stuff like that meet you at the box so people wanna call it test minister so wanna pewter who you I'd you can do that as well and they also almost want to mention is a POS improvement that's done by using the gold for library and then test for portal for that so it's makes it a lot faster to aggregate verify signature to be for to live verify 128 public keys

about 800 milliseconds and now we're able to cut down to 20 milliseconds for that so yeah that's it very cool thank you let's move on to lodestar hey so past few weeks we got a new full-time person on our team Eric weather weather cam on the call he's he's gonna be researching assembly script integrations he's so far already got an a shock 256 little annotation that he's kind of ported over to typescript playing around with just optimizing that on some other

fronts we're working on getting up-to-date with 8:08 to spec tests so we're rewriting our spec test loader it's kind of in in the process we've migrated our code management to a kind of a model repo style it's all in the lodestar repository and with that we've started breaking out some of our kind of sub modules or sub components so we've probably published SSC types on NPM we published the beacon chain configurations I've made that a minimal publish that so people can could use that they wouldn't like and then kind of

up next we're working on the networking for Interop we're partway through disc v5 implementation and the coming weeks we'll fix up our lid p2p protocols there's some kind of work on on gossip sub but a lot you know mostly on the request response side cool thanks next up is surely actually is way here okay Nimbus hi so we have a new joiner is in the call

Dmitri who we work on the leaf p2p and networking otherwise in the past three weeks we got a new SSD implemented on f1 we started about six months ago an equivalent to web Frito pie and web fridge is in name and now we are using it to interact with if one deposit contract we also have created a way to bootstrap if to Genesis from ganache so everyone can use it it's only joyous it's not mean I put the link in the chat

and on the networking side we implemented the new networking inspect and we also have live p2p working again on Windows and it can be tested we just met one thanks thank you I believe that was all the clients correct cool so proto has an update he's I think been chatting with white block and they're gonna be putting it together a survey to gather some information so program yes

and so we're closely working together on like to book and so what they wanted to do is really something like toaster key to map out exactly where from where you want to go and since this comparison table for all different features of course there's so many already many different features we know there's also a few features we have no Reese's certain circumstances and then he also discusses bring all the debugging features and what really needs to money garnishing and then from their own with

like some miss Dietrich memos like a short enough to look like the first playing songs in short yes together like a table like you know comparison chart that allows us to identify where each team is that so you can figure out what priority needs to be when we get together Peter ah and also just on drop engineer also almost these reasons for doing so sponsors documentation they're also working or said I'm going Africa must be absolutely and so it's really useful to see it burns effective to just let me just read yes so we'll be reaching out to

different client teams to try to get with a questionnaire or survey essentially to fill up and we'll probably put that on my github and share that and also try to like reach out one-on-one with everybody you know we're gonna speak with some of the other guys while we're in town like this everyone else that's here so anyways in Berlin let's grab a beer drunk cool thank you I know when I was talking with Raul a couple weeks ago he had intended to do something similar for the you're gonna handle that in the survey document that will give everybody

an understanding nicely what was that less yes agreed yeah thanks guys for that together hey Danny just to jump in for a second rolls on holiday it's a public holiday in Spain but yeah I think it would be good good to get in touch with them about this he has a lot of ideas about Interop and how we're gonna test interrupts between different limb p2p implementations so I think it'll be

working would be worth reaching out or I can make the intro or whatever yeah I'll follow up with Hennessy on next week yeah he should be on next week yeah oh cool and we'll be it ether linen if any guys around nice alright thanks Mike ok moving on to the next thing in our agenda which is networking since Felix is here Felix would you like to give us an update on discovery b5 there's also a number of implementations in the works at this point could you speak closer to the mic sorry

is it better this way so yeah so we started an audit of the spectra with least authority and it has been going on for one week they have created an initial report but so far the audit has mostly focused on the academy portion of the protocol and we are hoping to like the autumn will still go on for another one or two weeks at least but right now we're just at the point where I have to

go back and just write about the topic system more so that they will be able to evaluate better so basically it's a bit of like what this this first week has shown is that the thing is just not very well explained you know not so happy about that but um I will just explaining better and hope that it will work out and yeah on the implementation side I can say that not much has changed like my invitations basically up to the point where the spec is and I'd be very interested to actually see if we can do any kind of interesting on it with any

other implementation basically so if there is anyone is interested in that yeah we're interested in testing out the intro last last I saw there was some discrepancies between the go in the spec so they they were saying now yeah mostly because I changed this place so basically when yeah when I when I drop all the these calls the last update I gave was basically that the implementation was a bit ahead so now they are equal I think and we made the exchange because I wanted them to be part of the audit so you know I don't

know okay well update as to the latest spec and I'd be interested to the deficit is really minimal especially for the wire for the pole so there was one where in the the biggest change that that means that was efficient for me then the goal was that the tag prefix didn't gone from the who are you response packet and that basically that was a bit for me that was a bit weird like in the goal but it turned out that they actually can it revealed like some other issue in the implementation so but if your implementation like heavily relies on like parsing the tag out and then fussing around and basically it will be it will take maybe right now we're longer to fix it because the track

is now missing from one of the fact thanks just the first 30 mm yeah sure okay maybe I'll pick you up offline we can I can borrow some guy code and may I have a question to Felix yes absolutely the spec has a security consideration in the topic discovery section could you please give an update on that what do you mean I mean that spec says that adversary can send the wrong information about topics or something like that did I get it wrong so I don't know so

basically right now so the text is still there we just moved this part to the rationale so we did something like there's actually just a lot a lot of new text so when you last check this back what we actually added quite a bit of text before the audit to explain some of the security guarantees data that exists and those that don't exist in general it is not possible to verify if you are querying a single node for publishers or good I'd like for know it's advertising given topic you cannot be sure that the answer is correct because there is no way to verify that you know the notes you get back actually do provide that

topic the protocol cannot give you this fundamental guarantee because there is no way to verify but if you carry a lot of notes for providers of a given topic and in general the answer is probably gonna be okay thank you thank you it's been a couple of weeks ago when I check this back yeah just like there's a bunch of new fancier accounts cool thanks Hey hmm what do mean Iceland's icons little graphics the arrow hand and the the tools every five spec I just want to that there are icons

there are green and yellow icons representing different parts of the text wait are you checking why are you checking Oh Don it seems to be on Oh oh this is very no and you should check the main the actual content of the repo yeah sorry I just a quick google search method yeah no problem no most of the text is actually like all of the text so far is at least but it's part of the report yeah most of the now like the big change compared to a couple weeks ago now is

that basically we updated the rationals and out it matches the wire protocol specs so before there was the rationale was like stuck in January and the wire protocol was from April so now they both up to date cool okay any other questions for Felix for LaVon great Mike from protocol atmosphere Mike do anything for us today um yeah no role and I really didn't didn't have much to update about I mean I think everyone knows about the networking spec that was merged I was like 10 days ago we we're also starting on bounties project we're gonna kick it

off a teeth Berlin if any if you want to get in on that it's in a repo called lib p2p slash dev grants EEE grants but yeah we didn't we didn't really have anything else for this week yep and these bounties are going to be at the hackathon are these just more in general yeah so there will be bounties at the hackathon the organizers of eath Berlin why chose us as keep like a key project to focus the bounty pool on but some of these some of these projects were posting are like too complex to do in a

hackathon so you know we would be interested in other people applying and we're also hoping that other community members eventually might start posting their own you can just post an issue in this repo and then that's about D so might start posting their own and even started offering burns on the or somebody are they gonna call had mentioned maybe shot on you had mentioned that there was some work towards integrating noise yes that experimental I know the spec is still not quite defined yeah the noise implementation and JBM that p2p is there it kind of works but it's not complete

and that somebody can actually integrate it into their protocol yet it's just a couple of tests that show notice works with live other PDP i'm honey on making it more robust well and that's something i know Raul and others that are interested or kind of prioritizing suspecting that out in the coming weeks or month yeah no that's like I think we're all his priority list and noise was either number number one or number two so yeah it's it's it's definitely on our radar and at the important yeah cool thanks Mike great white block gel amazing that [Music]

[Music] we've been doing this week we're summarizing into documents that we'll share and strictly your short leaves it's been reviewed parties so brother something on the next comments could be more foremost talking sháá great any other networking things before

we discussed some of the attestation aggregation stuff is on the internet yeah I have a very brief question is that this v5 mentions handshake and I noticed that there was some potential overlap with the noise handshake so I just wanted to find out the best way to chat about it asynchronously so I'm essentially asking how to get in touch with Felix asynchronously already using the secrets created by the handshake but

I think it is maybe the best idea to reuse these or anything yes yeah yeah I agree it's the integration of the other secrets into the handshake which is where I see the overlap coming in in the interval additional data yeah so it's definitely like something to consider for the future that you know if since discovery also already has a secure and authenticated Channel might be the best to just add some some kind of message into the discovery already where you can say hey I want to open a transport

connections with you can also be very helpful for things like networks also because if you can you might actually be able to like punch a hole in an app or something with this kind of message or this is not part of the protocol just but it's definitely something that can be done at any time like anyone yeah this aspect we can edit sounds good okay thank you great okay so next up I Erickson is the last name what's your first name anyway

some of the harmony guys have been doing a dive into network load with respect to a cessation egg aggregation after station propagation you know if there are essentially multiple agent chain blocks and multiple shard blocks that are up there different network hues at the time we might actually have more a station load than expected and then also investigating the pros and cons of various strategies the very structure strategies the unstructured strategies and they have a good write-up of some of the problems at hand in the github issue I'll share it with you as how Candide I

but and I guess the conclusion is that we might need to increase the max expectations for blocks if we're if we have difficulties in aggregating such that we have some subsets of aggregation of vegetation aggregates that overlap with others and if we have multiple targets for beacon blocks and share about decisions I think that is a possibility certainly it comes at the expense of larger beacon chain blocks and essentially at the expense of computation you can likely still verify

the same number signatures with the trick that Vitalik created but you still have to do more summing of the public keys there's a lot of trade-offs and design here that's definitely being explored one thing that metallic suggested this morning is and to avoid some of the aggregate overlaps we can actually have smaller shard subnets so for example a committee split across two shard subnets such that the best effort aggregate from each can be broadcast and a proposer can accept it and then aggregate it fully without worrying about overlap so

there's there's a lot of little trade-offs laudable design decisions that we need to both think about benchmark and kind of explore on more real world networks thanks Danny yes I can speak a little bit about that yeah we have started our small research from investigating aggregation strategies and specifically in structured aggregation strategies and after some time we realized that probably the aggregation strategy is not is not a bottleneck or

for the system and mmm the station delivery could be a it could become a bottleneck especially when we will have big a big number of validator nodes like it could be starting from phase one or phase two or whatever and what we are these documents is just draft which contains some thoughts and some finished solutions that we want to share with everybody and wanna put these thoughts and problematics into it everybody's

head and after that get a feedback mmm and to prove that assumptions that made during that research are held and maybe after that we will have reasoning to continue to investigating that problem to do some modeling so yeah the basic one of the basic problem we have thought about is the DFT and the scalability which sure could which like um it's like there is a

contradiction here if we have a for example at ten or fifteen thousand nodes subscribe to beacon attestation topic we will probably have to to to low the degree of a p2p graph to make dissemination done in a limited amount of time but reducing degree of the graph will reduce its Byzantine fault tolerance property as well so you wanna read about that in these documents and just give us some thoughts on that or

whatever also there is a what compression question we we might want to increase max attestation number that could be included into a block there is a number of reasons to that and Alex also have thoughts about how we could compress the other stations in the block by a restructure in a block and we would probably need to compare this compression to two generic algorithms like snappy and we will probably do this

in the future so but yeah this this now is dropped for that so yeah that's it maybe Alex cares any other comments I think you explained clearly for now so I've had some other issues but I think it's enough for now it's it's it's a huge topic actually and it requires some time to get involved in that and we'd like more teams to be involved in that and also researchers thanks Danny for they can look at all

this stuff I hope we can get some feedback later on yeah absolutely so I actually before the zero three release I will put in in the valid in respect what the aggregation strategy is for interrupt and leave an opening form a net as we continue to explore these concepts so that we at least have like the kind of naive strategy that everyone can work on for Interop yeah it might be that nave strategy or will even work for phase zero for the main net right so

yeah and I expect that we could discuss these things in real life during the Interop I hope I think it will be useful for everybody yeah absolutely yeah cool so kind of as mentioned before the trade offs and some of the subtleties here are kind of one of the big four phase zero the big things that we need to figure out and also figure out what strategies might be valuable to pursue in the future as the network size grows cool okay thanks my girl is there any any other questions or comments on

how to stationary ation in this general area of research the pp Network great okay I skipped research so we're moving back to research any other any research updates Metallica some no sir and so one thing is that a couple of cars got merged so one of them is 1186 which is modifying the way that merkel proofs work order them working the morkul proof your

vacation works and part of that is that it includes basically instead of separately providing an index and a depth that it provides a generalized index and there's a set of methods for working with a generalized and this is there's also Oh an unfinished that PR which is I think it's it's either thirteen sixteen or 1323 it's the beacon chain them updates work for phase one where basically we get it adds in the episode persistent committee roots and then after that there is an updated or a PR that I started working on the Justin

recently made a bunch of changes view for making making light clients or defining the default light client a sinking algorithm so basically and they think the Fed they're in the important message is that there is now enough to go off of in order to your start design actually building light clients hold on there is a question in I'm sorry what was the PR the change tomorrow for verification format I'm one one eight six as you can just go to the document that's called I

think markup for updates or something like that or Merkel proof verification yeah so he thing then right so this basically includes it includes generalized indices which the other parts of the face once back are starting to directly use it includes Merkel proof single item in the momentum will be multiple item of your item verification and then let's see if I can extract the the PR for actually yeah yeah like what the like client side

changes its one the one three one six is the white coin sink 1 3 2 3 is the phase is the phase one beacon changes so yeah and it's it's there and I did encourage young clients and developers to start looking at it and also think about how to start implementing it one thing I also wanted to talk about in this regard is just the pasta the possibility of making some kind of infrastructure to make it easier for people to start building in the testing the roid clients like basically the challenge being bed

with the way the like clients back is currently is set up it's not going to be really possible to make a practical like client until phase one actually actually launches and so while we're learning out phase zero in parallel we wants to do all of this phase one stuff also in parallel trying to get fellow white clients of running as much as possible so I mean once did your own suggestion I had made is just the possibility of making some kind of minimal and of centralized test that that has one key one computer and friends runs one shard that you can have RPC RPC against to get to get charred black roots but would be

interesting if it think of some different options mmm-hmm because the quilt team is building out part at least the you charge a dirty data chains of Phase one in lighthouse we might in a few months have a good base client if we expose the methods work against as well okay perfect so you'll need to talk to will about that hey will update okay and the other thing that I just and have started quietly looking at is private information retrieval protocols to try to make light clients more

privacy-preserving and so far it looks like stuff exists but nobody pretty much increases the overhead significantly yeah meaning I can make a request to a server and the server doesn't know it doesn't know what it's the ideas basically that you make requests to a server in the server replies of a server has no idea [Music] [Music]

what you and whiteboy we talked about like one protocol a total sink we talked about the club private in private information rich people things yet [Music] conversation yeah on can you do the private information if you move out trust execution environments okay so there's three kinds of API are just to give us a summary here one kinds just to get the easy one is is basically using trusted execution environments the second kind is information-theoretic PIR

and the way information theoretic PIR works is you basically talk to end server and then you you ask on each of the servers to evaluate a function on all of the all of the data where in each server gets a different function and these functions could just be like a linear function so a vector that you dot product the entire data with but you choose these functions in such a way that the sum of all of the function equals to the function that just basically outputs the case P piece of data so like zero times everything but the KP is plus one times the capias and then you receive all of the answers from the server's you add up the answers and

you get exactly what you want and so DB from an information if you're writing point of view and none of the server's know it's like what you're actually accessing and the only way they could figure it out is if all of them colluded so the idea here would be that this would be based on some kind of - minority assumption where say you assume that at least five percent of registered servers are honest you asked twenty of them and then you add up the answer is if the answers from the twenty another alternative is and then the third alternative is computational PIR and computational PIR basically works by sending the server like some kind of

mathematical objects like could be a prime number mod eats you or some weird group elements like there's a bunch of protocols and then the server applies some transformation based on the based on every piece of data and the mathematical structure has some trapdoor that you include that basically has the effect that without the server's knowledge when you do when you have decrypt the results at the end it only includes information about the specific piece of piece of data that you wanted to know about so there there's different ways to do this but in the basic like the main drawback as you can tell over the second and third kind is basically

the server side there is this om n computational overhead and in some cases there's some some data overhead you can generally make it be a lesson though sure and another update on the white plants stuff I know I checked him with salts and there he's moving forward with integrating Raiden into the Catholic client framework and is still pushing on that he said stop buried so hopefully we see some cool stuff a lot of that in the next month or so wants to have a demo

for Def Con so I have a few updates as well one is that I guess starting from right now we have phase zero bounties so any substance to the phase zero spec station isn't function folk toys rule or deposit contract will get rewarded five if or thousand died so I guess a thousand day right now is more than five if so you just need to submit an issue or PR you know explaining the

substantive change or the problem that needs to be fixed and one that once that is merged in before the e through Genesis you'll get the bounty another update is BLS standardization so it's now kind of feature complete and there's no known issues so I guess we can consider it frozen in a similar way that phase zero is frozen so I guess the the plan is to integrate the new BLS spec into dev but not merging merged into

master too soon in particular we want we don't want to kind of disrupt the interoperability effort but at the same time we do want to incur at least some subset of the implementers to start playing with the new BLS standard I guess most of the effort will be implementing this new constant time hash function why be Bonet the rest is very kind of minor and and easy to to implement if you already have the currently at a spec implemented so just

to give it a little bit of context on the the other standardization it's part of the IETF standardization effort and actually it's spread across three different standards and none of these standards are fully ratified so there is the possibility that bugs kind of will be seen and fixed I guess the good news here is that a firm is going to be the first to launch I mean we still have a few months before launch so there is room to fix bugs if necessary you know at least until the deposit contract is launched so how do you find do you go

find these bugs I guess related to their destination there's also standardization around you know key related stuff that Carl has been driving so checksums HD wallets key stores we kind of want standardization around all these things for sanity and I think the approach that car has been has been taking and it's one that I quite like is somewhat minimalist design is already been discussed but I kind of want to share like different viewpoints I think the the new daikon design is is totally awesome so just I mean it's not it's not like a canonical like lion design I'd

say it's like a minimal one and there's there's a whole design space around like clients but at least the minimal one it's extremely simple you know on the order of 64 lines of code extremely low go ahead for the beacon chain because it's based on these persistent committees which don't change very much and it's also awesome because it has lower bandwidth requirements than in Bitcoin ste and you know I'm also investigating another optimization which could you know potentially in practice need to a 10x reduction in bandwidth so you know I'm personally quite excited

about the electrons once we have these these ultra light clients that can be used for more than just you know user facing clients in mobile phones they can also be used built to build these bridges between other block chains where gas gas cost optimization is critical and then the final thing that I've been looking into is this idea of a quantum apocalypse insurance so the idea here is that the validators can commit to to two secrets so in the context of a lamp or

signature which is quantum secure when when they make a deposit and we we can basically in in the event that there's some sort of quantum apocalypse quantum computers break pls signatures well we can basically disable and all the BLS signatures and have a way for for validators to safely transfer the balance as validator onto you whatever new platform comes next which which is quite insecure and the nice thing here is that we can do it without any

substantive change to phase zero and you can kind of think of it as a as a validator level convention so we we will be you know enforcing at the consensus layer these this this infrastructure but if you want to benefit from it then you need to kind of follow the convention there again swabbies but it is a substantive change is that it adds a field correct so in it it only adds a field to the transfers and the transfers are not part of phase zero so there's no substantive change there thanks Justin well would you like to give us an update

yeah for sure so a few things so we've been doing work on benchmarking basically single pass multi branch updates multi normal proofs and a sparse normal tree implementing implementing ssz within an execution environment in the context of simple transfers so that's been working on that and the eyz team has been doing a patrician Merkle tree with rfp and so we're getting preliminary results benchmarking numbers

on that so hope to have Morgan filled by in Iraq regarding all of that also we are continuing to build out phase one to prepare basically being able to plug in scout and B capability and phase 2 into to a client so that's that's in action right now we're doing that in White House we have most things kind of in place we're just kind of working on tedious pieces right now and yeah I'm filling in code and some bugs as well

let's see we started work on a new repo which wraps to some rapping on scout in particular and just basically gives you one execute function with a priest a truant input and an output and then to basically extract away I love it the FFI functions they're set up so we plan on just plugging that straight into into our our node that we're working on yeah that's that's a big equal great thanks to any questions for will cool Musab is here with 1/2 verification give me

updates for us yeah so we have we are well so we are reaching a state where we have more complete case specification of the phase zero specs so hopefully by the end of this week we will have something there this is going to be followed by body essentially building the testing framework and making sure that all the tests asked for and also updating the the specs to the point eight point two

right so that's that's good yeah cool and the other project which is the deposit contract I was talking today June two days ago and it's generally complete other than the a couple of negative conditions that he wanted to verify along with this Viper compiler error bug and the checksum that we're going to add but it's exciting generally verified and you ready to deploy when we're ready pull any other research updates great okay

general spec discussion general Interop discussion anything want to address over on this call together you do you mean by interrupts they mean that lock in you know you can give us an update yeah okay yeah sure so III know originally said kind of be there at 3:30 and I think what we're gonna wind up doing it's kind of doing waves so like kind of bands leaving to take us out to the you know the cult headquarters and so I think I'm gonna set up or I'm gonna add the pertinent

people to the there's an interrupt telegram channel and then tomorrow I'll do a call like QA about the sort of stuff because this could take a while but generally the idea is a sixth root of 13 you don't need to bring food its cater we have a cabin that supports like 30 people wasn't team has also gotten a cabin that's ports like and issue people and so we also happen with extra space that's that's what I was talking to you is talking is that the same day II was um or he wasn't Kevin yeah we got two different kinds great okay so yeah so

there's there's there are those cabins and so anyway so will provide food we order food for about 45 people so even if you're staying in another cabin it'd be great that I knew about you so you didn't starve to death because that would be cool and so yeah Antwan now going to have to the technical organization because I'm kind of blocked down with operations so he's going to help kind of organize the technical portion of what we're doing the idea though is that we're working towards interrupts in between our

clients because tuning and if you have any particular accommodation requests like Johnny ray asked her like a feather top you know king bed and bowl full of red gummy bears great cool they're excited yeah we'll take any questions and further discussion on this outside its call thanks Jersey cool general spec comments questions any anything that's come up with implementing some a new tests proto you got something for participating in

distance and if you lunch this general there's a little space extra space related directly to that is Benson for outsiders to work on and to completely bathroom to them and maybe even like bloom turn on Sierra scientist for me okay if I think a little bit more and getting people actually I gave the dust nuts trying to the school of the business this especially said this all the extreme cases then tried to reamer them in some part of and a fee or people are interested

yeah so there's there's still an effort to figure out a good way to do an NFC for participating in the deposit contract especially maybe early participants and I've been talking to Austin Griffith a little bit about that so yeah live in addition the test net said I I agree I I think we even intend to do some incentivize test nets such that we have particular bounties to break certain types of things so maybe we should start organizing ideas around that I like NF T's as well maybe even just achievement of like making X percent you know things like things just

ask questions like go over in real life make some fast service and just message discards the Geezer photos and yeah cool yeah you mentioned fuzzing actually I haven't heard any updates on fuzzing and like a month or so has anybody been in touch with the posing efforts designing of note to report there yeah maybe here I'm with Adrian so we've just been fuzzing lighthouse for panics and updated offices to the latest spec in

that process we actually caught a bug in our SS that implementation so that was a good one but I'm not so sure it's happening with the differential fuzzum yeah so the differential fuzzing has kind of gone quiet I guess I think proto maybe you can update on the latest status of serenity but from what I understand now it's all been so we go suspect is still over there that already that is is he going temptations to eat authority and big smart and so it's actually the first is even patient right now so every tuning for is right there it just has

been like like relatively quiet from Quito and then there's no like new developments in fishing but yeah make sure spec are working and then nice house where the next target great yes that's good okay anything else guys

okay last item on the agenda open discussion any question any comment anything that's not even quite related cool okay we'll call for the day I believe two weeks time we'll meet and then about a week after we will be gathering together in person thanks everyone thanks to her buddy buddy guys thank you [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music] [Music]

[Music] you

[Music] [Music]

[Music] [Music]

[Music] [Music]

[Music] [Music]

[Music] [Music] [Music]

[Music]