stokes:Okay, thank you.
Transcript
stokes:Hey everyone, welcome to ACDC number 174.
stokes:Let's see, this is issue 1907 in the PM repo.
stokes:Let me drop a link to the agenda there.
stokes:In the chat.
stokes:And yeah, so today, a few things. There's an interfork update, as we've been calling it.
stokes:Around cell-level deltas, which are useful at the networking layer, and tying to blob scaling.
stokes:I did want to check in on…
stokes:EPBS with Glamsterdam, and to get a sense of, yeah, how the dead notes are going there, and where we think we're at.
stokes:And, yesterday was the deadline for submitting headliner proposals. We now move to the phase of discussing these headliners for Hagota.
stokes:And making progress on selecting one. So…
stokes:Well, let's go ahead and get into it.
stokes:First up, yeah, I just try to sequence, sequence these in terms of work order, and given that this topic is kind of in between, say, Fusaka and Glamsterdam, we'll get into cell-level telpas.
stokes:Marco wanted to give a brief update on that, and yeah, I'll probably just turn it over to him now.
Marco Munizaga:Right off the bat, oh man.
Marco Munizaga:I mean, give me one second to share my screen.
Marco Munizaga:Okay.
stokes:Yeah, no worries.
Marco Munizaga:Well, I don't have, like, too much prepared here, but I just wanted to give folks a…
Marco Munizaga:Fairly lightweight update on, the cell-level deltas, and, you know, just kind of…
Marco Munizaga:For visibility, if you're not aware of it…
Marco Munizaga:Let me share my screen.
Marco Munizaga:Feel free… Here with questions.
Marco Munizaga:And, yeah, I'll… the right thing… Okay, share… Alright.
Marco Munizaga:Can folks see that alright?
stokes:Yep.
Marco Munizaga:Okay, I'll zoom in unless someone says something.
Marco Munizaga:So, right.
Marco Munizaga:Sell all the others, here's some useful links, and I'll share this, markdown later.
Marco Munizaga:But the idea is, instead of pushing full data columns to peers, that we're doing today for, the…
Marco Munizaga:DA destination phase.
Marco Munizaga:We can instead exchange bitmaps and only exchange the missing cells that appear as missing, and this is better in cases where we have at least some blobs that are present in the mempool, which today, this is all cases.
Marco Munizaga:And it's a backwards compatible optimization. You can enable and disable it with a feature flag.
Marco Munizaga:This is just… Use transparently whenever the two peers over Gossip sub support it.
Marco Munizaga:And, yeah, to give you a sense of how far along we are on implementation, I asked these folks to just give a quick update on the status.
Marco Munizaga:So, it doesn't have to be big, but, you know, just like a quick update, just randomly start with Daniel.
Daniel Knopik:Alright, so from Lighthouse side, we are almost done with implementation. Only, recent change to the specs have not been implemented yet.
Daniel Knopik:But otherwise, it's just a bit, you know, general polish left, like some better logging metrics, etc. But otherwise, close to done.
Marco Munizaga:Very cool. Arsh?
ashah:Yeah, hi, sure, so…
ashah:Thanks, thanks, Daniel. So on the Prysm side, we are reviewing the PR that,
ashah:marker raised to implement the partial cells, work in Prysm. I think we've been making good progress on the review. Now, there were some missing bits in there.
ashah:Around handling, eager push for partial data column headers. That's what we are, working on right now, and we need to incorporate some of the latest COSSIP sub changes around sending request bit masks.
ashah:On top of the parts metadata.
ashah:And after that, we plan to do some really solid testing of this end-to-end. But yeah, almost there.
Marco Munizaga:Thanks, Arsh.
Marco Munizaga:Dustin, are you here? Do you wanna… Favorite word?
Dustin:It's short.
Dustin:Yeah, sure. So… We have, our, our WebP2P implementation is, is in…
Dustin:testing, and so we also have some ongoing effort, sort of ongoing work on the, sort of, the CL side, sort of for brief reference, and Nimbuses are… or for IFT slash Nimbuses are two separate organizations doing the P2P and the Ethereum client.
Dustin:Currently, and so…
Dustin:That hasn't been… so we… we are not yet at the point where we can promise a specific shipping time on that, because that has…
Dustin:God.
Dustin:Yeah, it's part of the… it's… the P2P doesn't support it quite yet, but or it's the… the full end-to-end testing on the P2P side is not there yet.
Dustin:And, hopefully once… once that's there, we can… Kind of… Move reasonably quickly, but…
Marco Munizaga:Cool.
Marco Munizaga:Thanks, Dustin.
Marco Munizaga:Any questions here? Pause for a second, just to give folks…
Marco Munizaga:I know this is very, like, brief, so if there's any.
stokes:It sounds like… it sounds like Matt, might have an update for Lodestar as well.
Marco Munizaga:Oh, sorry. Let me open the chat.
Marco Munizaga:Oh, yeah, Matt, sorry.
Matthew Keil:Hi, yeah, we also have a tentative implementation, going through some debugging and just cleaning up, but, getting closer as well.
Marco Munizaga:Thank you, Matt.
Marco Munizaga:Am I cutting out? I'm sorry. It's a little choppy, but I think we can make it work.
Marco Munizaga:So, the next thing that I just want to, like, highlight is we have some… So, just rough…
Marco Munizaga:thoughts on how we're gonna test this. We have, like, a…
Marco Munizaga:Blob DevNet, thanks to Barnabas for spinning this up.
Marco Munizaga:and as we finish implementing the consensus specs here, we have some test cases we want to cover.
Marco Munizaga:I mentioned this here not as, like, a super exhaustive list, but, like, also to try to get some input into, if anyone here has
Marco Munizaga:Ideas that you think are important for this that is not covered.
Marco Munizaga:So, like, you know, basic cases of, like, oh, everyone supports partial messages, or…
Marco Munizaga:This is all private blobs, and there's nothing in the mempool, so we should see similar…
Marco Munizaga:Patterns as, like, the full data columns.
Marco Munizaga:a case of, like, you know, there's always one cell missing, some large portion of… or some large portion of the network does not support partial messages, and we should… we should observe that peers are still well and randomly connected, so we're not… we don't see clustering, we don't see peers, like, automatically prefer,
Marco Munizaga:Peers that support this optimization?
Marco Munizaga:And just a couple of other things you can read there.
Marco Munizaga:And then, just… Briefly talking about possible deployment strategies, so…
Marco Munizaga:I think what's cool about this optimization is
Marco Munizaga:we can enable it, and we can disable it. So, we'll do a ton of testing in the DevNets and testnets, and this is where I did not finish writing the slide. We can…
Marco Munizaga:progressively roll this out to validators we control, and kind of observe how they do over time. And if we ever notice anything weird, we can always just
Marco Munizaga:remove that feature flag. And, yeah, this isn't something that requires
Marco Munizaga:like, the whole network to adopt at once. We… we can, you know.
Marco Munizaga:Do this as progressively as we feel comfortable with.
Marco Munizaga:That's it, that's all I got. Any questions here?
stokes:Thanks, Marco.
stokes:Oh yeah, Barnabas?
Barnabas:Yeah, I would like to raise that, you know, Blob DevNet Zero has been running now for.
Barnabas:two weeks. It's been very stable, luckily. We've been trying to collect some metrics, to still get the lab to catch up. Maybe in two weeks I can give a better update, but, we're gonna be collecting some information about,
Barnabas:they get these three metrics, and we also have a few metrics in Graf Island dashboards, that Marco has put together.
Barnabas:They're still a work in progress, but hopefully we're gonna be able to see a lot more details in a couple of weeks.
Barnabas:Ideally, also, all the new metrics should be implemented by all the different SEA clients. If you have any questions regarding those, let us know.
Marco Munizaga:Thanks, Barnabis. I'll link the PR that talks about which metrics we're adding.
stokes:Okay.
stokes:Yeah, and then I guess, in terms of, yeah, getting this to mainnet,
stokes:So it is really nice that we can do this progressively. And, you know, the idea here is that this can be done in between forks, you know, it is a backwards compatible feature.
stokes:So, in terms of deployment, you know, I think this is something we should try to keep pushing on. I don't think anyone is rushing to get to another BPO, you know, in the immediate term.
stokes:But, as I understand, based on, yeah, the design and initial testing, this is actually, like, a huge bandwidth savings on the network, so I think the sooner we can get it out, the better.
stokes:Yeah, so… Just to have on everyone's radar.
stokes:Do any client teams see any issues with, you know, moving forward on this?
stokes:And what I mean concretely is, like, deploying this even before Amsterdam, you know, assuming we get things ready in time.
stokes:Because if not, that would be the proposal to move forward here.
stokes:Even before we think about more BPOs.
Marco Munizaga:Correct, yeah.
stokes:Yeah.
Barnabas:Yeah, I think we should draw it out,
Barnabas:As soon as we can, and maybe also, the CIP validators could enable these, so we actually see a few,
Barnabas:A small portion of the mainnet network,
Barnabas:Having these partial cells enabled on them before we… Make, like, a mass release.
stokes:Right. Yeah.
stokes:And I think the main risk here, again, because it is backwards compatible, is just, like, some weird issue where nodes would kind of tend to cluster on the network.
stokes:And that's, I think, worth calling out, just implement carefully, and I think Marco
stokes:can help with any questions there. But yeah, I think there's some notion of peer scoring that might tend to this, correct, Marco?
Marco Munizaga:Correct, yeah. This is kind of like a implementation pitfall, nothing with the protocol, but just a thing to keep in mind when implementing this, that you don't want to have a way for these partial message-capable peers to…
Marco Munizaga:Do better in peer scoring than their counterparts.
Marco Munizaga:I'll add one more thing that I forgot to add to the slides, which is we have a working group on this work. If you want to join, just reach out to me, and I'll add you to…
Marco Munizaga:the Telegram group, and we have weekly meetings just about issues that have come up and things we're working on.
Marco Munizaga:you don't have to join the meetings. If you don't want to, there's a…
Marco Munizaga:Doc that highlights the output of all the meetings and the current state of
Marco Munizaga:the work. So, if you want to see that doc, also feel free to reach out to me.
stokes:Okay, thank you.
stokes:Then, next up, we'll turn to Amsterdam.
stokes:So, yeah, for here, I did want to check in briefly, how we feel about DevNet Zero.
stokes:Yeah, last time, people were still pretty deep into initial implementations.
stokes:And, yeah. Any, any updates here worth touching on?
stokes:I do think in terms of keeping things moving, again, like we said, it'd be nice to have the DevNet Zero live sometime this month, so yeah, that's…
stokes:Something like 1, 2… Possibly 3 weeks from today.
stokes:How are people feeling about it?
stokes:Potus and Prism says we're targeting it.
stokes:Okay, so there is a new specs release, Justin, do you want to say a bit about that?
Justin Traglia:Sure. I mean, this pretty much just, moves the blob KCG commitments from the
Justin Traglia:Wherever to the, bid. So it's a networking change, or a data structure change, which should simplify data availability checks, and I think it's good to include in DevNet Zero.
Justin Traglia:That's the main change.
stokes:Got it.
stokes:So yeah, everyone, please target Alpha 2.
stokes:And… yeah, another update from Lodestar, targeting of the month. So, yeah, I think this should be what we aim for.
stokes:And, you know, until we get there, I think it's gonna be, yeah, tricky to think about other Cluster Dam EIPs, or, yeah, speak more about the timing.
stokes:of Amsterdam, more generally, so…
stokes:Keep up the good work, and… yeah, I guess on that note, has anyone run into anything worth surfacing here? Any questions about the spec, or, like, implementation issues that we should discuss?
stokes:Lotus?
Potuz:Yeah, there was a serious bug that was catched by the Lido researchers. We have one fix, but it's not clear that it's the best way of fixing it. It's regarding deposits. They encourage other clients to look into the current PR. This is in Discord.
Potuz:There's, there's another thing that I wanted to highlight, which is… I don't know if other clients, have, have been… have started implementing this.
Potuz:But I, I've been keeping, an annotated version of, of Glamstone, of FOCIL.
Potuz:And the last document I put out was from 4Choice, because I was myself starting to implement it. And then I realized that most of the complication in 4Choice, at least for Prism, doesn't come from Glowus itself, or it doesn't come from the 4Choice part, it comes from OptimisticSync.
stokes:Like, recovering from errors and removing nodes becomes much more complicated, and it has a lot of edge cases.
Potuz:So I would urge people that are starting to work on Forges, if you do this thing of, like.
Potuz:keeping a tree in four choice, and then suddenly your engine tells you that a whole branch was invalid, removing it is highly non-trivial.
stokes:Okay, yeah, that's a good flag.
stokes:I was looking at this PR. I guess on the for-choice stuff, like, yeah, we have, at least some infrastructure for four-choice tests. Do you think that that would be helpful for what you were talking about, BOTUS?
Potuz:Yeah, I actually haven't been following what, the status of the new fortress for Gloss is, because, I don't know if Misha is on the call, Terrence was…
Potuz:was collaborating with Misha, with Mikhail Kailini, in trying to use, this new format for the purchase test, but I don't know what the status of those is. But yeah, so we do have
Potuz:test for optimistic sync and removing branches, and even removing, like, jumping branches, in which, if you remove a branch that is invalid, and then your new head is also invalid, that we react correctly.
Potuz:So yes, if we keep those testes, it'll be fine.
stokes:Okay, I'll try to take a look there and, yeah, make sure that we have that going.
stokes:My sense is that most people are starting with, yeah, just sort of the core changes, you know, breaking the block into two, changes with this builder concept, and I think we're gonna probably…
stokes:wait, possibly even until DevNet 1, until there's, like, heavy fork-based testing, but yeah, that means…
stokes:It's time now to get the testing in place so we're ready.
stokes:When we get to DevNet 1.
stokes:Okay.
stokes:And yeah, I guess any other comments on this deposit bug with the 4897PR?
stokes:Any surprises there?
Justin Traglia:Potus mentioned something about it being a potential DOS factor, because of all the signature, verifications, I believe.
Potuz:Oh, no, I think it's very expensive to DOS on this. I mean, you need to pay to set that deposit. It'll cost you at least one ETH, so I don't think it's, I don't think it's so bad. I think cashing it is fine.
Potuz:Also, it shouldn't affect DevNet, because we're not going to have external builders in DevNet Zero, so…
Potuz:It's fine if we put it or not.
Justin Traglia:I should probably just merge the PR, then. It has a few reviews.
Justin Traglia:I'll do that after the call.
stokes:And I can take a look after the call too, Justin.
stokes:Cool.
stokes:End put us, just to clarify, when you say no external builders, the idea with .NET Zero is that everyone's just self-building?
Potuz:Yeah, so what we want to test on DevNet Zero is the separation of the payload and the block.
Potuz:We want to have the minimal DevNet that actually shows that we can handle that separation, both broadcast, catching up when we missed a payload, requesting it.
Potuz:Look, everything that is needed just to have a different block, a consensus block from the payload of broadcasting. So that means self-building, but still broadcasting them separately.
Potuz:And after we have that, then adding the builder and all of those paths, I believe it becomes much easier.
stokes:Yeah, that… that makes sense to me.
stokes:Cool.
stokes:Any other questions or comments on the state of EPVS right now?
stokes:Okay.
stokes:Then, yeah, I think then we'll keep moving all these threads forward, and yeah, everyone, please do your best to target DeadNet Zero again in the next few weeks.
stokes:It'd be really good to have something here by the end of the month.
stokes:And yeah, if you run into issues with… as you're implementing, yeah, please flag them on Discord, or yeah, you know, an appropriate venue, and yeah, we'll keep things moving along.
stokes:Raul, did you want to say anything on this link here?
Raúl Kripalani:Not specifically, just wanted to… to send it over here. Already posted it on underscore just to make it more discoverable.
Raúl Kripalani:But, but yeah, the idea is to, so we started, basically, this tool, takes, real blocks, SSZ encoded, and performs the transformation.
Raúl Kripalani:It does the bifurcation into the beacon block and the execution payload, and then has a breakdown of, all fields, the compression rate, and so on. And the idea is to enhance this with, as I said in chat, with BALS, but also, like, to make it a generally useful tool for us to.
Raúl Kripalani:To perform different kinds of analysis, specifically with regards to deadlines and variable PTC and so on. So it's just, just a tool for us, to, to, to use, to further, like, refine the fork.
Raúl Kripalani:So yeah, so just reach out to me if you've got any feedback, or if you'd like to see something specific here.
stokes:Okay, great. Yeah, it looks really nice.
stokes:One quick question, like, it has a delta when we do make the separation into the two parts of the block.
stokes:And it shows, you know, the overhead on the wire.
stokes:And most of them are fairly small, but there's a few blocks every so often where it's, like, you know, plus 1 kilobyte, plus 1 kilobyte. Do you have a sense of what's going on there?
Raúl Kripalani:Yeah, so I think what's happening here is that the snappy compressor is having a different table for once you separate the block, so basically the efficiency will just vary, so I think that's basically what we're seeing here. Yeah. I see. Interesting.
stokes:Hmm.
Raúl Kripalani:And it's just tiny fluctuations, really, so I don't expect that… like, this is kind of, like, something that I personally wanted to have, just wanted to know if we expected to see, like, any major difference on the wire, and I think it's negligible, so it comforts me to see… to know that it's this way.
stokes:Yeah, yeah, definitely.
stokes:Justin had a question in the chat. Why are the raw wire sizes different between food… well, yeah.
stokes:I think that was what we were just touching on, Justin.
Raúl Kripalani:Yeah, exactly. It's just that the compression rate, just varies on the compression table that the snappy… that Snappy is deriving for each payload, because we do… so there is some nuance here, because we don't do block compression, on the wire, we actually do the full…
Raúl Kripalani:Sorry, we don't do the streaming compression, the framing compression for Snappy, we do a block compression, so it just takes the entire payload and compresses it at once. That's, like, some nuance that's relevant here.
Justin Traglia:Okay, I see, makes sense.
stokes:Thanks for sharing.
stokes:Okay, then, yeah, if there's nothing else on Glamster Dam…
stokes:The next agenda item is EGOTA.
stokes:So, again, just to play back, the process here. So, we had a deadline yesterday for headliner proposals.
stokes:And now we will turn to discussion and finalization.
stokes:And the way that this works…
stokes:Let me just make sure I have this correct. Yeah, we said we'd have essentially till the end of this month, to do this, and the idea here is, yeah, getting community feedback, discussing them on ACD, C, or E.
stokes:And then ultimately finalizing the headliner selection.
stokes:So… That's the high-level process, and if we look at the headliners that we have,
stokes:You can see them here.
stokes:There's this tag, Hagota, on Ethereum Magicians.
stokes:And we see here that we have a proposal for…
stokes:I'm just gonna call out the CL ones. We have one for FOCIL, that we all know very well at this point.
stokes:I believe the next few are eel only. There's 2 new ones since I looked yesterday.
stokes:One is for a Lucid encrypted mempool, but I think that's an EL headliner.
stokes:And then one here for partial reconstruction and 2D PROS.
stokes:So, yeah, I think in terms of moving forward then, So I Spoke did want to, say something about FOCIL.
stokes:So, yeah, I think I'll go ahead and turn that over to you, and then we will…
stokes:Go from there.
soispoke:Yeah, thanks. Yeah, I'll keep it very, very brief, because we all know FOCIL by now, but I wanted to basically just make it official that we propose FOCIL as a Seattle headliner candidate for Hagota.
soispoke:Yeah, maybe I won't go back into what it brings, but just briefly, it, you know, significantly improves transaction inclusion guarantees.
soispoke:And it does this by enabling the multiple validators to build inclusion lists and participate in block building. Primary benefit is clearly censorship resistance,
soispoke:And again, transaction inclusion guarantees, and it benefits basically everyone that's transacting on-chain, so from everyday users to L2s to institutions.
soispoke:It does also benefit validators, you know, because…
soispoke:it, like, first of all, is actually a step towards enabling validators to contribute to preserving censorship resistance without having to be a local builder and sort of, like, sacrificed MEV rewards.
soispoke:And at this point, I think it's also kind of clear that
soispoke:ProSol is, like, a crucial component in the future architecture of the protocol, given the sort of, like, scaling and ZKVM directions we are headed towards. And lastly, in terms of technical readiness, you know, the design and specs have been reviewed and improved on for, like, 2 years now.
soispoke:I think today we have 8 out of 11 clients that have implemented For Soul prototypes. Of course, at, like, various stages, but, it's still quite a lot.
soispoke:And process-wise, it is, currently CFI'd for Hagota. I think it's the only one because of the…
soispoke:sort of, like, governance phase we had, for Glamstadam at the time. And, you know, if it ends up being SFI, the immediate plan will be to rebase on top of EPBS, and Gloss more generally, since I think the specs have been frozen.
soispoke:And yeah, that's where we are right now.
stokes:Thanks.
stokes:So, yeah, in terms of moving forward, I do see Leo here on the call.
stokes:Would you like to say something about your post to propose partial reconstruction in 2D PyRDOS?
Leo:Hello? Can you hear me?
stokes:Yep.
Leo:Yep. Yes, I would like to say a couple of words. So the idea here is, to do, some preliminary steps towards, full two-dimensional yardas.
Leo:I want to make clear that I'm not looking at a full two-dimensional
Leo:and complete vertical extension, I think we can start, with smaller steps, like just extend probably just a couple of rows.
Leo:So that the overhead in terms of, data to custody and bandwidth is not, too large. But I think, that will allow us to move forward into the two-dimensional implementation.
Leo:And also, I would like to propose to do the vertical extension outside the four-choice critical path.
Leo:So that we, you know, we… Have less risks into…
Leo:Into this implementation of 2D mission or PLAS.
Leo:So, yeah, take a look at the… at the proposal, but I think, in terms of moving forward for scalability, I think this is quite important, and this will allow us not only, you know, we already have kind of…
Leo:row reconstruction with, partial messaging and cell… I mean, cell level messaging and partial reconstruction, but I think it's important to have vertical reconstruction as well.
Leo:And, I think the… the way I suggested in the proposal is a possible way that is not very intrusive, is not very critical, and allow us to move forward towards collaboratively. So please take a look, and if you have any questions, remarks, please leave it in the post, or contact me directly.
Leo:Thanks.
stokes:Better? Thank you.
stokes:And… yeah. So… I had expected there to only be one headliner proposal, and I would have thought then today we could have optimistically made that selection.
stokes:But yeah, given that there's the second one here, I think it probably makes sense to give people some time to think about this, and then make the call on the next ACDC.
stokes:That being said, I think we should go ahead and discuss today, where people are at.
stokes:Following up from a comment that Ankara had in the chat, I do think there's broad consensus, with prioritizing CR for Hagota.
stokes:Yeah, so… How do clients feel about this? Or, yeah, generally anyone else on the call.
stokes:I do think, given… how we have been having the conversation around thinking about scoping for Hagota.
stokes:it seemed like FOCIL would be the headliner we would pick, but yeah, I'd like to open the floor for any discussion.
stokes:On Sigar?
Ansgar Dietrichs:Yeah, I just wanted to say from… from my side, because…
Ansgar Dietrichs:I mean, I'm both personally in favor of prioritizing censorship resistance, and I think fossil…
Ansgar Dietrichs:to me, looks like it's the right approach. I just wanted to basically say, like, in case we might even potentially make the decision today, or at least in two weeks.
Ansgar Dietrichs:Given that a headliner choice is quite the committal, decision, and I feel like there has been so much broad support around FOCIL that I think there's a bit of a risk that we just, you know, blindly kind of wave it through without one more kind of rigorous check of, like, are we, like, really 100% certain this is the right approach? Again, my intuition is the answer will be yes.
Ansgar Dietrichs:But I was just wondering, kind of, how many people here have already, like.
Ansgar Dietrichs:thought about this in enough depth that they have, like, very high confidence, versus is there something we should do, say, between now and two weeks from now, just to get to this 100% level of confidence? Again, might well be that people are already there, I just… I personally am not close enough to FOCIL that I'm necessarily yet fully there, but, yeah, just wanted to flag this.
stokes:Yeah, my senses… people are there, already. But, yeah, Matt.
Matthew Keil:We are strongly in favor of it, and support it, for Hegota.
stokes:Okay, great.
stokes:Any other client teams, want to speak to FOCIL right now?
stokes:I will echo then that people are strongly in support, as we were discussing this in Glamsterdam, so…
stokes:I expect that not to change, even if,
stokes:People aren't chiming in now? Okay. We got another comment from Teku, at least from Enrico on the Teku team, also.
stokes:Yeah. So that being said, we should follow the process, and given that, we kind of have this very recent other proposal with, the 2D reconstruction.
stokes:I think it does make sense, to take one more call, give people another two weeks to think about this a bit more. From there, also, yeah, in terms of ongoing concerns, just, yeah, taking a final pass and just, yeah, being…
stokes:Confident in our selection.
stokes:But yeah.
stokes:Because of that, then, I think that we should expect to make the headliner selection for Hagota on the CL side on next ACDC.
stokes:And then we can go from there.
stokes:I guess in terms of, Leo's proposal, is there anything, you know, like, should people just reach out to you directly, or do you want to…
stokes:Point people to a particular place if they have questions in the meantime.
Leo:Yes, feel free to, answer in the post, or reach out to me directly, or if anybody has comments right now.
Leo:Every question. Questions.
stokes:Okay.
stokes:Any questions or comments for Leo at the moment?
stokes:Okay.
stokes:Let's see, let me just make another pass on the agenda, so… yeah, I think that was it for today.
stokes:Again, I think people are pretty busy with, APBS.net Zero.
stokes:And… yeah. Otherwise, let's go ahead and be prepared.
stokes:To make the headliner selection on the CL side. Again, if we want to take some time on the next call, just to have, you know, more discussion.
stokes:We'll do that, but again, we're in a place to make that call two weeks from today, so let's do that.
stokes:And… otherwise… Yeah, anything else?
stokes:Okay, if not then, yeah, solid-level deltas to have, on… on the radar.
stokes:Of course, everyone's very heads down with EPBS step node 0.
stokes:Please, yeah, reach out if you run into any issues there, just so we can keep things moving.
stokes:And… yeah, then we'll make the call for the Hagota headliner on next ACDC.
stokes:And that should be it for today.
stokes:Thanks, everyone. I'll see you next time.
Potuz:Bye-bye.
Justin Traglia:Bye, everyone.
Ansgar Dietrichs:Everyone.
Chat Logs
00:07:23
stokes:https://github.com/ethereum/pm/issues/1907
00:09:16
Barnabas:is marco cutting out for anyone else?
00:09:26
Justin Traglia:https://github.com/ethereum/consensus-specs/pull/4558
00:09:27
Enrico Del Fante (tbenr):Replying to "is marco cutting out..."
Here too
00:09:27
Kevaundray Wedderburn:Replying to "is marco cutting out..."
A bit for me
00:09:39
Kalo | Obol:Replying to "is marco cutting out..."
Yap, for me as well
00:10:13
Raúl Kripalani:Replying to "is marco cutting out…"
Networking team could do with a better network 😂
00:10:41
Francesco:Replying to "is marco cutting out..."
Putting yourselves in the user’s shoes
00:11:01
Marius van der Wijden:Whooo Lighthouse lfg
00:11:01
Raúl Kripalani:Replying to "is marco cutting out…"
yep guys, this is how EIP-7870 actually feels
00:12:39
Matthew Keil:@Marco Munizaga I can also give an update
00:14:07
Barnabas:https://dora.blob-devnet-0.ethpandaops.io/blocks
00:15:06
Barnabas:Replying to "https://dora.blob-de..."
We currently have ~100% mev blocks
12 private blobs / 12 public blobs (can increase further)
We have 20 clients - 10 clients support partial messages
Everyone is up to head, everyone seems to have close to every other peers - full mesh
00:17:05
Marco Munizaga:https://github.com/ethereum/beacon-metrics/pull/21
00:17:58
Barnabas:I’d encourage everyone to have an opt in flag, not an opt out. And once well tested, enable it by default (cc lighthouse)
00:18:29
Justin Traglia:I support doing this asap.
00:21:05
Potuz:we're targetting it
00:21:06
Justin Traglia:We did publish a new release, with new specs for devnet-0: https://github.com/ethereum/consensus-specs/releases/tag/v1.7.0-alpha.2
00:21:41
NC:Lodestar is still targeting end of this month
00:22:48
Justin Traglia:Link to what potuz is talking about: https://github.com/ethereum/consensus-specs/pull/4897
00:23:52
Justin Traglia:Replying to "Link to what potuz i..."
Also see: https://discord.com/channels/595666850260713488/874767108809031740/1468323583548723282
00:24:08
Justin Traglia:Replying to "Link to what potuz i..."
I think caching deposit signature verifications is reasonable, but I could be wrong.
00:27:15
Raúl Kripalani:ICYMI, we just shipped this: https://gloas.ethp2p.dev simulates the separation, and will be updated soon to add BALs, and deadline analysis based on actual network activity
00:29:14
Justin Traglia:Replying to "ICYMI, we just shipp..."
This is super cool 🔥 one quick question, why are the raw/wire sizes different between fulu and gloas?
[Full message cannot be displayed on this version]
00:30:05
Justin Traglia:Replying to "ICYMI, we just shipp..."
For example, 240/238 for fulu and 240/235 for gloas.
00:30:38
Leo:It would be better to not normalize the bars, so that we can easily see the differences between slots
00:31:03
Satyajit:Replying to "ICYMI, we just shi..."
Looks great! An arrow to move forward/backward among slots easily would be great.
00:31:05
Potuz:@Francesco I want to merge this https://github.com/ethereum/consensus-specs/pull/4884 :)
00:31:50
stokes:https://ethereum-magicians.org/tag/hegota
00:32:01
Raúl Kripalani:Replying to "It would be better t…"
AFAIK the max bar width is the max of what’s displayed in the view (so it’s already relative). Are you seeing something different?
00:33:13
Enrico Del Fante (tbenr):Replying to "ICYMI, we just shipp..."
is ratio sign hardcoded to be - ?
[Full message cannot be displayed on this version]
00:34:20
Enrico Del Fante (tbenr):Replying to "ICYMI, we just shipp..."
Seems the first two should be positive, not negative
00:34:58
Ansgar Dietrichs:I think we have broad consensus on prioritizing CR in H*.
I am not super deep in the weeds on the specifics of the FOCIL mechanism - how much consensus is there around this specifically being the right approach? Is this something that needs more discussion? If so, would we do that before or after selecting it as headliner?
00:35:25
Ansgar Dietrichs:Replying to "I think we have broa..."
I think I remember some skepticism by @Toni Wahrstätter for example?
00:35:37
Potuz:Replying to "I think we have broa..."
I do not think there is any alternative mechanism proposed
00:35:53
Raúl Kripalani:Replying to "ICYMI, we just shipp…"
Hmm good find, this looks like a bug.
00:36:27
Raúl Kripalani:Replying to "ICYMI, we just shipp…"
Should be fixed shortly!
00:36:55
Ansgar Dietrichs:Replying to "I think we have broa..."
yes I just want to confirm we actually think it’s the right thing to do. My assumption is that the answer is yes, but important to be certain about this.
00:36:56
Julian Ma:Replying to "I think we have broa..."
Do you mean whether FOCIL as a whole is the right specific approach or whether some of the specifics of FOCIL should be changed like exact enforcement rules of the IL and size etc
00:37:18
Ansgar Dietrichs:Replying to "I think we have broa..."
@Julian Ma FOCIL as a whole - adjustments can always be made post headliner selection
00:37:25
soispoke:@Ansgar Dietrichs I don’t want to speak for Toni but I’d say some critics were mostly around prioritizing CR/FOCIL for H*
And maybe the conditional/unconditional property (but it’s more of a FOCIL parameter)
00:38:27
Toni Wahrstätter:Replying to "[Full message cannot be displayed on this version]"
Yeah agree, it feels like the majority is for focil and this is what we should then go for as a headliner 👍🏼
00:38:28
donnoh | L2BEAT:Replying to "I think we have broa…"
from my convos with optimistic rollups in the context of reducing the challenge period, focil is not seen as a blocker but rather other things mainly related to networking or unknown bugs (concerns mainly coming from offchain labs). i’d love to have a better economic model to understand benefits in terms of economic advantage, mainly in terms of additional cost to attacker to bribe to miss slots (or the IL committee). as a comparison, 6s slots increase the attacker cost by 2x, not sure if FOCIL is more or less
00:39:16
Toni Wahrstätter:Replying to "[Full message cannot be displayed on this version]"
Agree with ansgar here though
00:39:40
Julian Ma:Replying to "I think we have broa..."
I’m not aware of any other mechanism. The big property FOCIL does not give is CR for MEV txs.
The proposed mechanisms for that - multiple concurrent proposers, explored mostly for e.g. Solana - seem to be converging on being very very similar to the FOCIL architecture
00:39:41
soispoke:https://ethereum-magicians.org/t/hegota-headliner-proposal-focil-eip-7805/27604
In the proposal btw we mentioned the tradeoffs (and the critics) explicitly
00:39:43
DA | Flashbots:Replying to "I think we have broa..."
There’s quite of bit of consensus in MEV research that FOCIL is a pretty weak design. Worse than unconditional ILs and much worse than some other less fleshed out designs.
But the impact on builders is very little. So we’re not blocking.
00:39:44
Toni Wahrstätter:Replying to "[Full message cannot be displayed on this version]"
Should now also be considered in light of AA or encrypted mempools as potential EL headliners
00:40:07
Julian Ma:Replying to "I think we have broa..."
@DA | Flashbots which other less fleshed out designs do you refer to?
00:40:13
Enrico Del Fante (tbenr):FOCIL!
00:40:24
Ansgar Dietrichs:yes sounds like in the comments as well that people have high confidence in FOCIL the specific mechanism already
00:40:45
DA | Flashbots:Replying to "I think we have broa..."
@Julian Ma some form of splitting auctions into many auctions to have multiple builders construct single blocks.
00:41:36
Francesco:Replying to "I think we have broa..."
There’s less fleshed out and there’s “people have been talking about this stuff for years without anything implementable coming out of it”
00:41:54
Ansgar Dietrichs:I personally think we’d pretty clearly be in a place to make a decision today already, but also fine to wait 2 weeks
00:42:24
Leo:Replying to "It would be better..."
I meant the main bar, the ones below are not normalized, so they are good. But every time the main bar is refreshed we need to look at the number to understand. Is a detail, but I thought it could help.
00:42:34
Francesco:Replying to "I think we have broa..."
Imho we should very much consider having the design be unconditional for now, everything else can hopefully be built in the future from the foundation of multiple contributions to the building process
00:43:17
Julian Ma:Replying to "I think we have broa..."
From what I understand to be the current main proposal, it’s based on this minus the erasure coding part https://eprint.iacr.org/2025/1772
It’s basically FOCIL infra + some small rule changes + potentially encryption
That is to say, if we do FOCIL, we could extend it into MCP. FOCIL is a necessary step for it
00:43:19
nixo:bye!
00:43:39
DA | Flashbots:Replying to "I think we have broa..."
@Julian Ma let’s jam in our call next week
Summary
11 highlights
· 3 action itemsExperimental
Summary
11 highlights · 3 action itemsExperimentalinterfork optimizations
glamsterdam epbs
- EPBS DevNet Zero targeted for end of February by multiple clients00:20:39
- New specs release v1.7.0-alpha.2 moves blob KZG commitments to bid structure00:21:06
- Lido researchers found deposit bug; PR 4897 addresses signature verification DOS risk00:22:45
- Fork-choice complexity warning: OptimisticSync with invalid branch removal is highly non-trivial00:23:04
hegota headliner
- FOCIL proposed as Hegotá headliner; 8/11 clients have prototypes, currently CFI'd00:32:52
- New headliner proposal: Partial reconstruction and 2D PeerDAS (vertical extension)00:35:40
- Broad consensus on prioritizing censorship resistance for Hegotá fork00:38:27
- Headliner selection deferred to next ACDC to allow review of 2D proposal00:42:11