Ethereum Protocol Fellowship (EPF) Cohort 7 — Applications open until May 13

FOCIL Breakout #029

2026-02-24 Agenda: #1946 canonical JSON

Transcript

00:03:32
Jihoon:Alright, I think we can kick it off.
00:03:36
Jihoon:So, hello everyone, welcome to FOCIL Breakout 29.
00:03:41
Jihoon:First off, FOCIL has SFI for Pectra.
00:03:45
Jihoon:Thank you everyone who has been working on and supporting FOCIL.
00:03:50
Jihoon:In response to that, I've realized also spec on Gloas, and then promoted it to here. As he…
00:03:58
Jihoon:Here I was referring to CL spec, execution specs, and Engine APIs, or…
00:04:06
Jihoon:still outdated, but I don't expect any meaningful changes there, to be honest. I'm still discussing with spec maintainers.
00:04:16
Jihoon:On how to go about that rebasing.
00:04:20
Jihoon:Let me quickly share the link to the… as he's back.
00:04:33
Jihoon:Yeah, here's a spec, and… The gist of the spec change is… So…
00:04:42
Jihoon:let's say if you're in slot N, and when we receive a peak on block.
00:04:48
Jihoon:or near the attestation deadline for slot N, we call it get head, right? And if there was a payload for slot N minus 1,
00:04:57
Jihoon:We get to decide whether to extend that payload or not.
00:05:01
Jihoon:So any PBS, we rework a payload when half of PTC vote against it.
00:05:09
Jihoon:the beacon block for slot N was timely, and it chose to reorog the payload for slot N-1.
00:05:16
Jihoon:In here, we… let me fossil add a new condition that
00:05:22
Jihoon:If the payload was not IL compliant, we just reorgan.
00:05:27
Jihoon:So, with this condition, when a payload was not IL compliant, attesters don't vote for it, and a proposer doesn't extend it.
00:05:36
Jihoon:And yeah, that's the… I think that's the main change of the fossil spec. So, yep, that's it. That's for the spec update. Do we have any questions or any other updates?
00:06:16
Jihoon:Alright, thank… We don't have any questions.
00:06:22
Jihoon:Do we have any development updates?
00:06:30
Fabio Di Fabio:Not a real update, but, on a Besu.
00:06:35
Fabio Di Fabio:We are starting this week.
00:06:39
Fabio Di Fabio:Necessary activities for… integrating FOCIL to Besu.
00:06:47
Jihoon:Nice, will it gonna be… on top of… Glamsterdam? Or… Fusaka?
00:07:02
Fabio Di Fabio:It… On top of Glamsterdam, yes. Because we are…
00:07:09
Fabio Di Fabio:already many EIP for Glamsterdam that are our main, so… we usually… Right.
00:07:17
Fabio Di Fabio:To keep all the new features on Maine.
00:07:20
Fabio Di Fabio:As quickly as possible, and so you should be on top of that.
00:07:32
Jihoon:Okay, so… Yeah, it's really good to hear it.
00:07:35
Jihoon:And I think the next step forward Ahead of us, but be…
00:07:41
Jihoon:besides the spec updates, I think it will be a spec test, so that
00:07:45
Jihoon:We can provide test factors for clients.
00:07:50
Jihoon:So… Yeah, if we can have…
00:07:54
Jihoon:best prototype on top of Glamsterdam, then… I think it…
00:08:00
Jihoon:It will really come in handy when we test.
00:08:03
Jihoon:Possible. So, yeah, really good to hear it.
00:08:13
Jihoon:Okay, if we don't have any other updates, then…
00:08:19
Jihoon:We can go ahead and wrap up this call.
00:08:28
Jihoon:By the way, nico, we have a PBS implementation on Lodestar. Do you think…
00:08:37
Jihoon:Is it gonna make sense to…
00:08:40
Jihoon:implement fossil on top of it, or is it still premature?
00:08:45
nflaig:Right now, it's not yet possible, because we have not yet merged everything into our unstable. That's on the ePBS DevNet Zero. So, I think we need to clean that up first a bit, and get it, more reviewed and cleaned up.
00:09:02
nflaig:And then we can rebase our fossil branch against that, because it's based on unstable branch right now, not the ePBS devnet.
00:09:13
nflaig:But yeah, maybe, 2 weeks or something? Maybe we can try to rebase it on clause.
00:09:23
Jihoon:do we… wait for… Like, field deafness, or, like, deafness zero would be fine to rebase fossil.
00:09:35
nflaig:Yeah, I think the devnet zero scope, should cover everything for fossil. I don't think anything we add in later… oh, wait, maybe not. It doesn't cover the PTC, so we don't, want to implement that in devnet zero.
00:10:01
Jihoon:It's not… far out, so it would be, I don't know, a few weeks or 1 or 2 months.
00:10:11
nflaig:Yeah, I think at least from our side, we would like to have something, before the Interop. I think that's doable.
00:10:28
Jihoon:Okay, thanks for sharing the updates. If we don't have any other updates, then we can…
00:10:42
Jihoon:Alright, thank you guys, have a great rest of your day, and see you guys in 2 weeks!
00:10:49
Fabio Di Fabio:Thank you. Oh, bye.
00:10:51
Jihoon:Yeah, Palette did share something on the chat, I'm sorry I didn't see it.
00:10:56
Jihoon:Maybe, you wanna share something, Pelot?
00:11:00
Pelle Krabbenhoeft:Yeah, I can talk for a quick second. These two charts are just what I've seen so far using the block-native data, which has been a lot easier to work with, but basically…
00:11:11
Pelle Krabbenhoeft:The overview is with the top fee IL building, we're looking at about 85% for the zero slot delay, so the current implementation, and then…
00:11:23
Pelle Krabbenhoeft:We… Moving to the two-slot delay, it changes by about the same amount as I was talking last week. Let me just pull it up.
00:11:35
Pelle Krabbenhoeft:Yeah, you're going from 84% of the transactions in the inclusion list are included in the next spot, block already, and then with the delays, it gets up to 85.
00:11:47
Pelle Krabbenhoeft:So, not too much of a difference there. And… I found some more issues with this, so this is just a sneak peek, but, just wanted to show this off as well. And then, actual censored inclusion lists are also about the same as the
00:12:05
Pelle Krabbenhoeft:XL… XL data I shared last time, but the inclusion rates are a lot higher with this block native data, so I'm assuming… so I have some more testing to do, but we're looking at about 95%
00:12:18
Pelle Krabbenhoeft:of transactions included in these inclusion lists with the top fee, building structure.
00:12:26
Pelle Krabbenhoeft:Are essentially a waste of bandwidth, but yeah.
00:12:32
Jihoon:Okay, great. Yeah. Thanks for sharing.
00:12:38
Jihoon:So, I think the idea would be…
00:12:46
Jihoon:Sorry, I haven't read your document fully, but I guess the plan is to update the script to get more precise data?
00:12:57
Pelle Krabbenhoeft:Yeah, there was an issue with the config, and just due to the volume of all the data, it's taking my laptop a couple days to process, so, it's kind of just waiting on that, really.
00:13:19
Jihoon:Alright, I think we can really go ahead and wrap this call. Alright, thank you guys. Stay in 2 weeks. Thank you.

Chat Logs

00:04:33
Jihoon:https://ethereum.github.io/consensus-specs/specs/heze/beacon-chain/
00:09:23
Pelle Krabbenhoeft:Just some data on IL building. It’s taking longer then expected to step through all of the data.

Summary

9 highlights · 2 action itemsExperimental

fork status and schedule

  • FOCIL has SFI'd for Hegotá fork00:03:45
  • CL spec updated to Heze; execution specs and Engine APIs pending rebase00:03:58

spec updates

  • New reorganization condition: non-IL-compliant payloads are rejected by attesters and proposers00:04:48
  • IL compliance check integrated into get_head logic at attestation deadline00:05:09

client updates

  • Besu starting FOCIL integration this week on top of Amsterdam00:06:35
  • Lodestar: EPBS DevNet Zero merge required before FOCIL rebase; targeting DevNet One with PTC00:09:02
  • Lodestar aims for FOCIL implementation before interop00:10:01

analysis updates

  • Block-native data shows 85% IL transaction inclusion with zero-slot delay; 95% with two-slot delay00:11:11
  • Top-fee IL building: ~95% of IL transactions already included, representing wasted bandwidth00:12:18

Action Items

  • Spec team: Develop FOCIL spec tests for client implementation validation00:07:35
  • Pelle Krabbenhoeft: Complete block-native data processing for refined IL analysis00:13:08