ACDT2026-09-14_096

AllCoreDevs - Testing #096

2026-09-14 6 decisions 360 transcript lines

Transcript

Follow along

Potuz
gm
Justin Traglia
gm
Marcos A. Maceo
gm
jochem-brouwer
gm gm
Pooja Ranjan
gm
jochem-brouwer
Heeft gereageerd op "gm" met 🙏
Justin Traglia
Agenda: https://github.com/ethereum/pm/issues/2217
Daniel Lehrner
Fine for Besu
milen | Erigon
yes
Marius Van Der Wijden (M)
Sgtm
Luca | Vero
not super important for Sepolia but full Beacon API support a few weeks ahead of Hoodi would be nice
Pawan Dhananjay
Replying to "not super important ..." What’s not supported?
Luca | Vero
Replying to "not super important ..." POST produceBlockV4, SSE events, v2 proposer duties … it varies by client
Pawan Dhananjay
Replying to "not super important ..." Will double check, thanks!
Nico Flaig
sync also seems ot take forever for EL, is that related to Jochem's testing?
Luca | Vero
Replying to "not super important ..." there’s also this endpoint still being discussed - https://github.com/ethereum/beacon-APIs/pull/627 . Feel free to reach out, I can give more details
Pawan Dhananjay
Reacted to "there’s also this endpoint still being discussed - https://github.com/ethereum/beacon-APIs/pull/627 . Feel free to reach out, I can give more details" with 👍
jochem-brouwer
Antwoord verzenden naar "sync also seems ot..." I do not inject any new state (except required prestate to execute the attack, which was done ~2 weeks ago). So state should not be a problem regarding the attack scenarios
jochem-brouwer
Antwoord verzenden naar "sync also seems ot..." (this specific scenario, there is of course more network load)
pk910
Can we jump back to shadow forks for a moment?
Pawan Dhananjay
Maybe wait one more acdt before doing it? Given we need to handle a few bugs on lighthouse
Dragan Rakita
Leave a heads up in discord so people can see it
Jihoon
FYI, here are discussion links to which EIPs may add tests: CL spec issue: https://github.com/ethereum/consensus-specs/issues/5618 EthR&D thread: https://discord.com/channels/595666850260713488/1548032466487611392/1548032471914778644
Justin Traglia
Reacted to "FYI, here are discussion links to which EIPs may add tests: CL spec issue: https://github.com/ethereum/consensus-specs/issues/5618 EthR&D thread: https://discord.com/channels/595666850260713488/1548032466487611392/1548032471914778644" with 🙏
jochem-brouwer
RE: CFI to add tests, this itself is fine to me to end up in the trunk branch of the repository, but it should not be a signal to not work on tests when an EIP is in PFI stage. The tests should ideally be prepared, so once the EIP moves to CFI then it can directly be merged to the repo trunk branch
Jihoon
Replying to "RE: CFI to add tests..." Agree. People can put tests in a PR or their forked repo.
Justin Traglia
Reacted to "Agree. People can put tests in a PR or their forked repo." with 👍
felipe
Reacted to "Agree. People can ..." with 👍
Jun Song
Reacted to "Agree. People can put tests in a PR or their forked repo." with 👍
jochem-brouwer
Heeft gereageerd op "Agree. People can ..." met 👍
Enrico Del Fante (tbenr)
Replying to "RE: CFI to add tests..." I agree. It is important to have early tests for clients
jochem-brouwer
Heeft gereageerd op "I agree. It is imp..." met 👍
Justin Traglia
CL breakout: https://ethereumfoundation.zoom.us/j/89441658268?pwd=yjqf4TheEb1R7ZSuesCdOAy5tdyVk7.1
spencer
Reacted to "Agree. People can ..." with 👍
Fabio Di Fabio
haven't looked at it yet
Maria Silva
https://github.com/ethereum/devp2p/pull/281
Maria Silva
Restore pre-EIP-7610 legacy test expectations legacytests#18
jochem-brouwer
There is also this one, what is the difference? https://github.com/ethereum/legacytests/pull/17
milen | Erigon
Replying to "There is also this o..." The first reverts to Yellow Paper + EIP-684
milen | Erigon
Replying to "There is also this o..." These tests were existing prior to EIP-7610
milen | Erigon
Replying to "There is also this o..." The 2nd removes them completely
jochem-brouwer
Antwoord verzenden naar "There is also this..." For clarity which PR is first and which is 2nd?
milen | Erigon
Replying to "There is also this o..." 1st - https://github.com/ethereum/legacytests/pull/18
Karim Taam
Also for snapsync, we use it for healing
felipe
we could `fill_until(Osaka)`?
jochem-brouwer
Antwoord verzenden naar "we could `fill_unt..." Right so we mark the tests as unavailable in Amsterdam. I think this works
felipe
Reacted to "Right so we mark t..." with 👍
jochem-brouwer
I see
FLCL (Nethermind)
Maybe we can make a single fixture that people can test and compare then in the chat to be sure and close the topic
spencer
Reacted to "Maybe we can make ..." with 👍
spencer
Replying to "Maybe we can make ..." yeah lets do that!
spencer
Replying to "Maybe we can make ..." we can have this ready for tmo
FLCL (Nethermind)
Reacted to "we can have this ready for tmo" with ❤️
jochem-brouwer
This is a text change, not spec change, right?
Dragan Rakita
Reacted to "This is a text change, not spec change, right?" with 👍
FLCL (Nethermind)
Reacted to "This is a text change, not spec change, right?" with 👍
felipe
Reacted to "This is a text cha..." with 👍
spencer
Reacted to "This is a text cha..." with 👍
spencer
can try add a test for this now too!
jochem-brouwer
Heeft gereageerd op "can try add a test..." met 👍
jochem-brouwer
Thanks all :D
FLCL (Nethermind)
Ty!

Call summary

Targets

•October 6th - Sepolia Glamsterdam upgrade - 00:10:26
•Next week - official EL test release including EIP-8037 test - 00:50:49

Decisions

•Sepolia Glamsterdam upgrade scheduled for October 6th; all clients confirmed - 00:10:26
•Consensus-specs trunk restricted to CFI/SFI EIPs; PFI-stage tests developed in forks/PRs - 00:19:11
•Sepolia shadow fork delayed ~1 week pending Lighthouse progressive containers bug fixes - 00:18:24
•Merge legacytests#18 (revert to Yellow Paper + EIP-684 collision behavior); BAL collision left undefined pending EIP-8253 in Hegota - 00:35:09
•EIP-8037 merged as text clarification only; no spec or client behavior change - 00:49:28

Highlights

Devnet Updates:
- ·Glamsterdam devnet 8: Jochem's attack stopped; Besu and Erigon recovering, Nimbus resyncing - 00:06:32
- ·Glamsterdam devnet 11 launched last Wednesday; Glamsterdam activated cleanly, no participation drop - 00:07:48
- ·Lido and Optimism deployed on devnet 11; both confirmed smooth upgrade - 00:08:00
- ·Future adversarial tests (spam, malicious actors) to run on devnet 8, not devnet 11 - 00:09:09
Testing Policy:
- ·Proposal: restrict consensus-specs test additions to CFI/SFI EIPs only in trunk branch - 00:15:57
- ·Consensus: EIP champions should develop tests in forks/PRs during PFI stage, merge to trunk upon CFI - 00:19:32
El Breakout Eip8037:
- ·EIP-8037 (explicit total gas limit for state access) merged; text-only clarification, no behavior change - 00:49:28
- ·Spencer: official EL test release targeting next week; EIP-8037 test to be included - 00:50:49
Fork Status And Schedule:
- ·Sepolia Glamsterdam upgrade confirmed for October 6th; all clients gave thumbs-up - 00:10:26
- ·Teku (Enrico): concern about progressive list size handling; awaiting CL breakout signal - 00:11:12
- ·Sepolia shadow fork deferred ~1 week; Lighthouse needs progressive containers audit first - 00:18:24
El Breakout Bal Collision:
- ·Legacy tests failing due to EIP-7610 collision behavior; proposal to revert to Yellow Paper + EIP-684 semantics - 00:30:02
- ·Plan: merge milen's PR (legacytests#18) restoring pre-EIP-7610 tests; BAL collision behavior left undefined for now - 00:35:09
- ·Karim: BAL used for snap-sync healing; missing state modifications in BAL could produce invalid state - 00:43:57
- ·Decision: create single fixture for BAL collision scenario so clients can compare behavior and close topic - 00:46:17
El Breakout Sparse Blob Pool:
- ·Nethermind using payload-form length for sparse blob pool (ETH/72); all other clients use network-form length - 00:23:46
- ·Nethermind to update implementation to match other clients; PR review requested - 00:27:57

Action Items

•All EL teams - Review devp2p PR #281 (sparse blob pool length encoding) - 00:29:22
•spencer (EF testing) - Create single BAL collision fixture for cross-client comparison - 00:46:20

Key decisions

Sepolia Glamsterdam upgrade scheduled for October 6th; all clients confirmed
SFIUpgrade on Sepolia testnet set for Oct 6.
Consensus-specs trunk restricted to CFI/SFI EIPs; PFI-stage tests developed in forks/PRs
Limits test additions to CFI/SFI EIPs and moves PFI-stage test development to forks.
Sepolia shadow fork delayed ~1 week pending Lighthouse progressive containers bug fixes
Shadow fork postponed to allow Lighthouse audit.
Merge legacytests#18 reverting to Yellow Paper + EIP-684 collision behavior; BAL collision left undefined pending EIP-8253
IncludedRestores pre‑EIP‑7610 tests; BAL behavior pending future EIP.
EIP-8037 merged as text clarification only; no spec or client behavior change
IncludedAdds clarification without changing behavior.
Create single BAL collision fixture for cross-client comparison
Facilitates client comparison on BAL collision scenario.

EIPs discussed

EIP-684 EIP-8037 EIP-8253

Links

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.