eth_multicall Meeting, May 6, 2024

Transcript

so just a short update from our side would be that we finally managed to uh see everything needed to start a new wave of reviews and the reviews will start tomorrow as soon as I'll finish the update of the return data from the logs that Kar pointed out and um the we actually fixed on Friday the issue that we had on Return of the uh of the guess of of some of the guas values that are different due to the fact that we actually now allow our processor to

consume the minimum minimum guas fee and not set it to zero if it is not set explicitly so yeah that's a minute issue but it definitely changed many logs and this would remove it so yeah running five tests one L time before the next review and if you'll get any any like ideas of what shall be fixed in there any like specific tests to fix after we'll ship the new image please do post to me in DMS or on trailer so that you'll be able to like

make it in parallel to the review yeah that's it with us sorry did you say there's a new image now or are you talking about the it will be out tomorrow okay okay have you checked the tests that it's passing most of the tests or have you been able to run them uh yeah I'm checking them right now uh sometimes it's it's easy to understand sometimes it's not that clear for me but yeah I I'll come back to you when I'll get my understanding more clear but basically they should always match except the error message is

something that can differ but other than that they should be matching and but yeah some of them might be get bucks as well if there if you find discrepancy and then you believe that you are right yep discuss with about that's why trailer is much more useful for us than then just running tests blindly and yeah I had one more question for like this current iteration our understanding is that we use the setting from the configuration that would set the inter block times to 12 seconds am I correct with that yeah that's what we agre on last time

yeah did you remove the Phantom blocks yes haven't you seen it in your last like running of it well I didn't run them but yeah I can run them and see please do and then CA hey guys um yeah not so much uh for me I was busy with other work yeah my my goal is still to um remove Shadow blocks still have to do that

and um yeah seconds yeah and yes change the block time to 12 seconds I guess that's I can do that today it's easy straightforward yeah I I'm guessing that this week also will be a bit tight for me as I'm going to be traveling okay did I had something to say before I interrupted yeah nothing just that we would really appreciate if uh gu would also use 12 seconds when we we'll be testing

tomorrow yeah sure I can I can try to push that today thank you but yeah the Phantom blocks is one that will cause some differences but yeah not all the test are using Phantom blocks we will hope Trail will help us to synchronize across what is still wrong or not yeah do we have anything new that has come up I don't think we have no anything new maybe now we can find L spec but I don't know do we have any outstanding issues

to make spec not finalized that's how it has been many times that we think that it's finalized but it never is but yeah I guess now nobody has found any issues let's give it one last week figure out all the issues on trail and uh maybe somehow ask Community one more time on just like on Discord in a synchronous way maybe can we do something like that what well I think before we can get all the Test match then there can be something that you guys have we understood differently and then we had to discuss those that we had

to agree on how some of those cases are handled yeah but if we could like Po Community somehow to just like take a look at this spec think about it propose something if they have any ideas yeah so that they would also have time before like final reviews so get them never mind yeah was it clear for you how the erc2 transfers work that you get those figured out it is clearer but I'll I'll have to run it after I finalized tests that we fixed with Lush yeah I think the data field is is the quantity being moved and then you have

to have those uh in the logs you have to read those topics which are I think the three are always the same do you have by the way some other test that would be able to test not like this special log that we eject but uh some like more default logs that would just happen to be ejected bym so that I would be able to compare the thing that we are creating like manually is to the thing that was created automatically is if you could send to me this test in DM that would be perfect yeah I think there's

some yeah I can I those in the DM thank you so much alas welcome Hey guys nice to meet you I just um I joined the the calendar invite for the some like all cor devs I'm currently working at kak KVM I don't know if you know it we've been working with some mind people it's um Z km out of the stet ecosystem nice and I was looking at the your article on on mirror I wait who is dark Flo they're not here

I yeah it's mine and um and we're looking at how we could uh implement the the E multi one on kakarat client but I don't have so much context so maybe I'll just sit in and listen a bit nice to meet you Kari made beautiful 80ish I think five tests that allow to test nearly every piece of what is expected from E simulate to to like be produced so yeah there is uh I think we can somehow share the

details of how you could like take that hi note and look into it yeah the the blog post is somewhat old we also it used to be called multi call and now now simulate and was changed and then I'm talking about the Phantom blocks in that and but they decide to remove the Phantom blocks then there can be some other the differenes like that H so you're saying it's not called multic call V1 anymore it's called simulate yeah that's right Sim better here's in chat here I'll paste a link to the uh spec that we're working

on so that that should should be up to date within like a week like when we make decisions here on this call it gets updated usually by the end of the week and we haven't made any super recently okay nice we're currently working with a fork of five to pass the hive test we're passing all the EF test but is this already available the hi The Hive brench you mentioned it's a fork yeah there's a branch for them but it's it's not merged to The Hive yet okay but you can someone should be able to just pull that branch and test against this right yeah but if you're interested in the

development we also have a telegram group that you can join and we can invite you there if you're interested in more discussion about it then we can help you more when you're implementing stuff okay I guess this falls into I don't know if you've talk to many uh ZK World teams like um scoll or um um p on about this no we haven't but we have mentioned in the cev calls once but no one of those reached out us I don't think we have reached out for those we have been most reaching out the core deps of the eum protocol yeah I guess they're waiting for um the update

to like kind of hit get or and then for them to rebase or something I don't know but um I guess one problem we had in K is um the I don't know if you know it but we wrote the evm in Caro which is the the Z KSL of stet and so we have a approvable evm but then in some parts that are like solely in the client so this happened for tracing for example and for debug API it was harder to it was we had to rewrite it in Rost or like use the use The Primitives from RE to implement it because we usually we if you use tracing you don't need it to be proven so you shouldn't write it in um provable

language but um so my guess is that this multical this simulate sorry it simulate V1 is something that is going to be because it's it's only a simulation right it's only going to be in the client side and so my guess is that if we were to help we would probably help by implementing it in W I don't know if they've already implemented it they have not and you're correct this would be a client s side only and um I can't think of any reason why you need to prove it unless you want to do something like uh a like client sort of thing where you have someone else run the simulation you want to them to prove to you they actually ran a legit simulation against right block like you could do all that um but just if you're looking for just a consensus type stuff

um it's all non-consensus um so similar to eth call in the sense that it's off train it doesn't matter um if you trust the person executing the codee then which it could be you then you don't need to or anything yeah um as far as re I believe we reached out to them and they didn't express any problems with it or concerns um but I think it just wasn't on the high on the road map so they haven't started it yet and so um you might want to just verify with them to see if they've already started it maybe and just haven't talked to us before so you don't waste any work um but because I do know they are I'm pretty confident they're aware of it um but last I heard the him start anything okay nice so I think and ask wa cool yeah we um if you want to give me

your telegram handle I can invite you to our telegram group and if you have any questions all you're researching or if you decide to implement it we can that's a good place to ask thanks a lot give me 15 minutes to figure out how to invite someone on telegram do we have anything else everything else clear actually try that try that invite link see if that works for you yep excellent okay thanks a

lot I guess we can W it up if you don't have a else oh good for me thanks uh nothing for me nothing for me looking forward to nethermind and gu uh both having same output and we can move forward hopefully soon to getting this into the main line of both clients that'd be great yeah see y bye bye thanks nice to meet you guys you too