AllCoreDevs - Testing CL Breakout #100
Transcript
- Jihoon
- Barnabé Monnot
is it assumed that includers must execute the payload before making their IL?
- NC
Replying to "is it assumed that i..." currently the spec says try this, if not, can fall back to making IL before newPayload
- Jihoon
https://gist.github.com/jihoonsong/b8d8bbf862f9fd7cd4dbb89fd0d47ca2
- Barnabé Monnot
Reacted to "currently the spec says try this, if not, can fall back to making IL before newPayload" with 👍
- Justin Traglia
yep
- Barnabé Monnot
Replying to "is it assumed that i..." i’ll just make the point here 🙂 in the future with transactions carrying PQ signatures that require aggregation, I think we might want to have more asynchrony between what includers include and the blocks that are produced anyways. so over time I’d expect we’re less likely to want to block on the payload
- Barnabas
I mean thats a client bug
- Barnabas
exactly
- Barnabas
lmao
- Nico Flaig
wasn't nimbus shipping the stale config :D
- Justin Traglia
Reacted to "wasn't nimbus shipping the stale config :D" with 😅
- Dustin
for one release yes
- Barnabas
Reacted to "wasn't nimbus shipping the stale config :D" with 😅
- Dustin
then turns out another client was
- Dustin
we fixed our
- NC
Would this impact our decisions on whether to adopt 10s slot? That IL committee takes ~1.5s to create IL after newPayload call?
- Dustin
they didn't
- Dustin
we regarded it (properly) as a bug
- Barnabas
Replying to "they didn't" report a bug report? I’m sure they fix in their next release
- Barnabas
Replying to "they didn't" easy to make a mistake
- NC
PR sounds reasonable to me
- Dustin
Replying to "they didn't" they'd fixed it in trunk long before
- Dustin
Replying to "they didn't" they didn't seem to regard it as a priority. they knew/chose not to
- Dustin
Replying to "they didn't" anyway it was too late, we had to accomodate so don't care about that one anymore, it's done/gone
- Barnabé Monnot
my strawman would be attestation deadline 3s -> 2s slot duration moves to 10s keep the same duration between IL deadline and end of slot
- Dustin
Replying to "they didn't" want to avoid in future, thus comment here
- Barnabé Monnot
Replying to "my strawman would be..." need data 🙂
- Luca | Vero
Vero has a --ignore-spec-mismatch flag for this
- Nico Flaig
9 seconds is also fine, just merge it and we can tweak later
- Barnabas
next devnet then targeting beta 5?
- Justin Traglia
Replying to "next devnet then tar..." Yep.
- Barnabas
@james can we get prysm on devnet 1?
- Barnabas
how is focil going?
- Luca | Vero
- Justin Traglia
For reference, this is the PR: https://github.com/ethereum/beacon-APIs/pull/627
- Jun Song
Replying to "@james can we get pr..." Prysm: we’re thinking of participating from devnet 2
- Nico Flaig
@other clients, ptal
Call summary
Targets
- •Thursday — Beta 5 release with quick slots tests included - 00:14:49
- •Devnet 1 targets Beta 5; quick slots STF changes not required until following devnet - 00:13:05
Decisions
- •IL deadline raised from 8s to 9s (or 75% of slot duration); merge approved with no opposition - 00:12:07
- •Quick slots PR (EIP-8198) to be merged; Beta 5 spec release targeting Thursday - 00:12:14
Highlights
- Client Updates:
- ·Beta 4 released; Hoodi releases must comply — contains a breaking change for an edge case - 00:00:27
- ·Lodestar passing Beta 4; spec tests will catch non-compliant clients automatically - 00:01:02
- Beacon Api Gloas:
- ·New Beacon API endpoint (beacon-APIs PR #627) lets validator clients supply their own bids, addressing multi-node setup monitoring complexity introduced by Gloas bid selection moving into the CL - 00:15:30
- Focil Spec Updates:
- ·Proposal to raise IL deadline from 8s to 9s into slot; motivated by worst-case payload processing time (~1.4–4.8s for 200M gas blocks) - 00:01:31
- ·If slot duration reduces to 10s, IL deadline would sit at 7.5s (75%) — tentative, to be revisited with real data - 00:07:44
- ·Clients should only validate fork-specific configs when that fork is actually scheduled, to prevent mainnet interop issues - 00:10:40
- Quick Slots Eip8198:
- ·Quick slots PR (EIP-8198) to be merged after call; Beta 5 release targeting Thursday - 00:12:14
- ·Current spec tests do not exercise the slot duration change; clients could pass Beta 5 tests without implementing STF changes — tests need to be added before release - 00:13:09
Action Items
- •Jihoon - Add spec tests covering quick slots STF changes before Beta 5 release - 00:13:41
- •CL teams - Review Beacon API PR #627 (validator-supplied bids); thread in CL APIs Discord channel - 00:18:29
Key decisions
IL deadline raised from 8s to 9s (or 75% of slot duration); merge approved with no opposition
IncludedIncrease the inactivity leak deadline to improve processing time margins.Quick slots PR (EIP-8198) to be merged; Beta 5 spec release targeting Thursday
IncludedMerge quick slots changes and include them in the upcoming Beta 5 release.
EIPs discussed
AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.