All Core Devs - Testing #87

Transcript

Justin Traglia: Good morning again.

Mario Vega: Good morning.

Justin Traglia: I was able to, or apparently I'm co-host this time, which is cool to see. Like, via Z.

Mario Vega: Yes, yes, I can see myself as a host. I didn't even see that. When I joined. Nice.

Justin Traglia: So I can mute people now. Thankfully.

Mario Vega: Great. Hello, Ben.

Ben Adams: Hey.

Mario Vega: Hey, everyone, let's give a couple more minutes to get time to be able to join, and then we can start. Thank you.

Justin Florentine: Good morning testers.

Mario Vega: All right, I think we have enough people to get started. So I'll just… I'll just… I'll just get started now. Yes, welcome everyone to ACDT number 87. Today is July 13, 2026. We have a very light agenda today. Just basically, the main topics are Glamsterdam. DevNet 6 and DevNet 7 deployment. So yeah, let's just get started for. The Amsterdam Debenhams Debenhams 6, current status. I think, Stefan put a very comprehensive thread on Ethereum R&D Discord. So, Stefan, would you like to kick off the… The status for Glasgow Dynamics, please.

Stefan Starflinger: Yeah, sure. So, in general, we have a pretty good participation overall. I think it's generally around 80 plus percent, which is very nice to see. There were just some minor things that I noticed. Some nodes, couldn't recover from a fork. It seems Lodestar, Bazel, Lodestar, Erigon, they're… the notes got hung up, because I think the EL refused to update its head. I just restarted Lodestar Basel, and it seemed to recover, and I left Lodestar Erigon. How it is for now, if someone wants to look into that. And otherwise, it was just minor things. I think NimbusCL had an issue, but that's already been resolved now, and it's working pretty well. I think Grandine, we looked into some issue where, there was some P99

delay in block production, but it was mostly just their events that they were publishing a little bit late. And overall, I think it's going pretty well. for Glamsterdam 6. I think… Is there anything that anyone would want to say regarding DevNet 6? Otherwise, I think we can move on to, DevNet 7. Oh, right, I know. Yeah, can you just continue?

Mario Vega: Yeah, for sure. I just wanted to first mention maybe the releases for Grand Theft Auto VII. We have 7.2 on the EL side, and I don't know the number of the CL. But yeah, Justin, do you want to?

Justin Traglia: I was just going to say it was out for 12.

Mario Vega: Perfect, perfect. So yeah, yeah, I think I think the consumption of the of the releases is the main question at the moment. Stefan, do you have like. any idea of how clients are doing on this regard.

Stefan Starflinger: So regarding the tests on the EL side, clients are doing pretty well. I think Nethermind, Guest, and ethrex are passing most of the tests. And they're already working pretty well in Kurtosis networks locally. I'm not sure, maybe the other clients, Besu, Erigon, and Reth, could give an update how far they are. With the EL tests?

Daniel Lehrner: I can give an update for a bit. So I have a local branch where we are failing around. 40 tests or so. We are passing the majority. I think it's something with the repricing that is wrong. But I can push the changes and then we can try it in Kurtosis. Because the majority of tests were passing.

Dragan Rakita: From Red's side, I'm passing all the stated blockchain tests and need to review changes and propagate them, so probably we'll have branch tomorrow morning.

Stefan Starflinger: Erigon, is someone here?

Andrew Ashikhmin: Yeah, I'm here. But I haven't caught up with the team member who is working on Glam. So them Devnet 7. So sorry I don't have an update.

Stefan Starflinger: Okay, and I think Nimbus here as well.

Dustin Brody: Yeah, I mean that the. We should be… Passing most of them, I'm, I'm not certain of the exact. Situation.

Stefan Starflinger: Okay, I can look as well in more detail later. I think you're passing most of the consume engine tests and the consume RLP tests. I will have to look that up.

Dustin Brody: Yeah, I know the branch is not 100% ready, but it's reasonably close.

Stefan Starflinger: Also, by tomorrow.

Dustin Brody: Okay. One would hope.

Stefan Starflinger: And I think next, maybe on the CL side, I saw a question from Justin. How far? Clsr, if they could give an update. I think…

Mario Vega: Is it Deco…

Stefan Starflinger: Yeah, the.

Mario Vega: My house. Everyone, anyone, anybody else? Any updates about the seal test passing. For the client?

Dustin Brody: Nimbus passes all of the tests.

Justin Traglia: I know Lodestar does as This Prysm passing. Oh, cool. Personals password.

Terence Tsao: Yeah, or passing.

Justin Traglia: Is there, is there anyone here from Grandine? Yes, that's the only one left Okay, we can follow up on that later.

Stefan Starflinger: Regarding DevNet 7, I've been doing some local testing and found a few issues that I've reported. One about peer scoring with Prysm, then a P2P issue with Nimbus. And, otherwise, the else I've already reported, one Nethermind didn't connect to any outbound peers. But I think we're very close to launching. I think I've pretty much everything ready. Either, after the call, I will, launch it soon. Or maybe end of the day or tomorrow morning. So I'm planning to launch with a bigger state so we can train test the Soc. State, so I'm planning to launch with, half a million exited validators. With a validator balance of 1024 ETH withdrawal credentials of 0x02. And I'm planning to manually deploy the 8282 contract.

So that we can kind of test, how it would be. On the live network. That's kind of the update from my side for definite 7.

Mario Vega: There was one question from Dustin. Do you know which gas limit is?

Stefan Starflinger: I think right now I've set it to 200 million, but I'm also open to suggestions.

Mario Vega: And also, I guess, are you running any fuzzing? transactions, or… Spam, anyway.

Stefan Starflinger: Yes, we will start with… well, we will wait for the fork, and then we will be running fuzzing transactions, a mix of different ones.

Mario Vega: Ben asks for 300 million.

Stefan Starflinger: I think we could, if we all have, I think, E70 implemented. Then we should be okay with a higher gas limit. for the DevNet.

Mario Vega: Any other clients have any thoughts on 300 million? Maurice is sure. Any thoughts from Nethermind, Basu? Everyone.

Ben Adams: I think never mind that, yeah.

Mario Vega: Oh, yeah, that's necessary. Sorry Besu seems okay. Reth.

Dragan Rakita: Yeah.

Mario Vega: Okay, Stefan, are you okay with 300 million?

Stefan Starflinger: Yeah, let's do it.

Mario Vega: All right, let's go. Yeah. Cool. Any other questions regarding Amsterdam? That's it. PL or CL related. That is true. Do you want to expand on that? Should we also see a client for this?

Potuz: I just wanted to mention this, that discussions about gas limit should involve the CL as the CL is now a determining factor on the slot restructuring. And moreover, this is also something for CL devs. CL devs should know the numbers that benchmarkers are pointing because we should be taking part of this discussion. So we should be more informed than we typically are.

Ben Adams: I I hope you're on the call.

Potuz: Sorry, Ben?

Ben Adams: I said I'd hope you're on the call. Maybe you should run through the CL people.

Mario Vega: Yep. So apart from Prysm, are there any other CL clients that want to oppose 200 million?

Potuz: Wait, wait.

Mario Vega: Limit for.

Potuz: I didn't say that Prysm is opposing anything. We're happy with 300 million as well. I mean, I'm happy. I'm not sure about the other Prysm members.

Mario Vega: Got it, okay. So, yeah. Any other SEAL clients have any talks about either 300 million gas or 200 million gas? On the… on the glancer.devin7. All right. Yeah. Seems like it's a go. We stick with 300 million. Thanks for raising the point, potuz. All right. I think that's it for general updates. There's 1 extra topic that is not related to either El or Cl. That I want to bring up. That was put in the agenda today, which is the Glamsterdam upgrade mascot. Basically just go into the, I'm gonna share the link to the final decision. And, It's it's here. Basically, there's a decision. Now, please, clients comment. If there's like opposition to the final mascot for Amsterdam has been decided.

Basically, that's it. Regarding the rest of the topics, we're gonna have a different dynamic today, regarding the EL breakout and CL breakout. For the EL breakout, we're gonna stay in this same call, and for the CL breakout, we're gonna jump into a different call, right? Before we jump, are there any topics that have to be discussed by both EL or CL before we jump into the different calls? If not, then we can just proceed. Basically, ELDevs, stay here. CLDevs, Justin just shared the link to the other call in the chat. Please jump into the call and then we can start the discussion.

Justin Traglia: All right, bye all.

Mario Vega: All right. I guess we can just start with the CL. Sorry, EL. discussion topics. So the only discussion topic for today for the Eel breakout is this. share the link. Yeah, this update to EIP-8037. This was found by Jochem. And also… I mean, I guess the champion is Tony for this. Does Jochem or Tony want to share a little bit more context on this? update.

Jochem Brouwer: Yeah, I can share. Without this update, it's possible to construct blocks which are 8 times the size of what we actually would expect. This updates… Fixes the book where the floor cost for the data floor accounting is only done at the transaction level and not done at the block level. So this updates solves the problem that the flowing is not applied to the block size.

Mario Vega: Okay. This has been already mentioned. It was also on release 7.2, I guess. Does any client have any comments on this? Yeah. Tony, anything to add to this this point?

Toni Wahrstätter: Not really. I reviewed Jochem's PR, and it seemed like a clear bug. I did use PR. So please, I think it's… I'm not sure if it's already merged, I have to check, but please review that one, as well as Jochem's PR. I see Spencer saying it's merged, perfect. Yeah, it was a simple change.

Mario Vega: This is a breaking change, Stefan. So basically this is a consensus difference. We have it in this 7.2. release. As far as I know. So basically, if… yeah, it is included, basically, but I'm not sure we didn't have a chance to discuss this, basically. That's why I'm raising the point also. I don't know if there's any opposition to this in Devon 7, or is if there's like Any comments from clients, basically. That's mean. main issue.

Maria Silva: Oh, just… just to clarify, it's, like, it's already merged in the specs, so I guess, like, clients would have to change their implementations to see all this… passing. Is this correct?

Mario Vega: James, that is that is correct. Okay. Spencer, can you confirm this is 7.2, right?

Spencer Taylor-Brown: Yeah, so, I guess, yeah, so this was, like, mentioned on Friday. Just, like, a day after we did the 7.1 release, and yeah, it seemed like such an obvious, Spec change that Jochem found, so just tagged a release really quickly and got the 7.2 release out, so. And as far as tests breaking, I think it may be… Change the couple tests, and then… To Tony's PR, I added an extra test. to that before merging. So there's an extra test and a change to, like, I think two or three tests, so it shouldn't really affect And. Much, yeah. So I think it's a… it's a fair hot

But yeah, I guess relaying what Mario said, if anyone's against it, now is the time to. To see.

Mario Vega: All right. I think that's basically it for today's discussion for the EO. I mean, is there any other topics? I mean, just to close this one, does anybody have any questions about this? this EIP change. Yes, Maria, please go.

Maria Silva: I was going to raise another point, so let's first close this one.

Dragan Rakita: Just to repeat this about floor gas, basically getting included in the regular gas for the block.

Maria Silva: That is, that is correct, Ran.

Dragan Rakita: Okay, that's good.

Mario Vega: Yes, as far as I understand, it was, the now service. Potential issue. Any other questions around this, or can we continue to the next…

Jochem Brouwer: Yeah, I just want to correct that denial of service point, because it's not really a denial of service. It's about the points raised by EIP-7623. So, we have a max-sized block, at least we assume this, but before the spec change, we could actually produce much larger, larger box than, what we thought was our maximum. So it's not really a denial of service, but it's Okay, you could also see it as a denial of service, but it's more that our assumptions did not hold anymore.

Mario Vega: All right. I think we can continue. Maria, do you want to bring up the next topic? Hey, how are you, man.

Maria Silva: Yes, so this is mostly a question that I wanted to ask you. We are mostly at the… end of the of the Devnets. Right? And I was wondering what is the appetite for like doing some small tweaks that are not like they are not changing the mechanisms, but they are clearing, like, making the EIPs clear or something. So, what I mean, like, one of the ideas was That Johan was suggesting, for instance, instead of the regular gas variable in 1837, we could maybe call it something like execution gas, which is a bit more forward compatible as we go with more dimensions. It wouldn't change anything on, like, this… mechanism itself, but it would maybe require some renamings in the specs.

So I wonder, like, what is the appetite for, at this point, introducing these sort of, like, small changes for the next DevNet? Or do you feel like we shouldn't do it?

Mario Vega: Yeah, I.

Jochem Brouwer: Yeah, to add to this point, so these changes would only land in the EIPs. So for clients themselves, like if a client wants to call it a regular guess, execute guess, compute guess, it doesn't really matter. The thing is that these EIPs, this is the first thing that people are going to read about this fork because it's an official document. And this is also our, well, I would not say only, but our first and most important chance to actually create, well, yeah, like good terms for this so that this is force compatible and not confusing in the future. This is also why I think that they should. carefully consider, like, what is actually the text of these EIPs once we move them to final.

Because if we… introduce, like, I'm not saying that regular gas, for instance, is, like, a vague concept. But if you can just clarify it a little bit more so that the ecosystem, the community can understand it faster, then this is the time to do it. And we should not do that like in the next world that we say, okay, we are going to rename this because that would then require an EIP. to basically redact or to alter these EIPs. So, yeah, that's… that's why I'm proposing, like, to… to… yeah. To do that in a good way now.

Maria Silva: I think, other thing that I was thinking about, was, and I think someone raised this, I don't remember where, so apologies if I'm not, saying the name of the person, but, on 2780, we have this additional cost for the logs. And then we realized that actually there are some other logs that are emitted and they are not priced directly, like with calls, for instance. And, so maybe it would make sense for us, for now, to just keep everything consistent. So we wouldn't have a specific log transfer cost in 8037. And instead, we just have the base cost and the transfer with value. The values would be summed up to be the same. So we still have the 21K. It's just we wouldn't have these with value being as low as it is now. It would be a bit higher because we are not separating, making these log costs separate.

So I think it would actually make the EIP a bit more. a bit clearer, but it would mean, for instance, a slight change in the variables, values in the variable names. Not the variable names, but we would remove the log part of the name. of it. So… so yeah, How do you? How do people feel about doing this on Amsterdam? Devnet 8. That be feasible. Okay. If yes, I can… make the edits to the EIPs, and then share. Okay.

Dragan Rakita: I think it's desirable. It's going to simplify how the people are going to read it, so I think we should do it. I would not, 2780 could be… There's a lot of modifications that we could additionally do to that IP, and I would restrict ourselves to, like. Not doing more consensus, bigger consensus changes, and go with this, like, simpler approach. But renaming, doing more, like, cleaning up, everything around that seems… Really great.

Maria Silva: Okay, perfect. So I guess. Yeah, that's all I wanted to ask. Thank you for your feedback.

Mario Vega: Are you going to propose the PRs and then share it with the… That's right.

Maria Silva: Yes, yes, so I'll make those changes. I don't know, I will make them this week, but I don't think there's also rush, because we are still working on DevNet 7, right?

Mario Vega: I mean, from what I understood, these are not going to modify any consensus rules, right? This is just basically just clean up, right? So I don't think it affects us in any way.

Maria Silva: Okay. Okay, but yeah, so I will try to do it, not sure if I'll be able to have it this week, but at least, next week should be…

Mario Vega: Sounds good. Alright. Thanks. Thanks, Maria, for sharing this. I think this is basically the last topic that we have for today. So I wanted to open quickly the floor to anyone that has any other topics that want to discuss regarding basically anything related to Devnets or any other topics. Alright, going once… Twice. Price. Cool. Yeah, I guess that's it for today. Thanks, everyone. This was a quick call, basically. Thanks, everyone, for joining, and catch you on Thursday call. Thank you.

Jochem Brouwer: Thank you. Bye-bye.

Toni Wahrstätter: Thank you, bye bye.

Gary Rong: That would be nice.

Transcript adapted from ethereum/pm, CC BY-SA 3.0.