ACDT2026-06-15_083

AllCoreDevs - Testing #083

2026-06-15 2 decisions 359 transcript lines

Transcript

At a glance

Correctness over speed: EIP-8282 will be included in Glamsterdam; formal ratification at ACDE Thursday
EIP-8282 targeted for glamsterdam-devnet-6 alongside remaining repricing EIPs; devnet-6 target next Wednesday
Test vectors for devnet-6 EIPs — Spencer targeting Monday June 23
glamsterdam-devnet-6 launch — targeting Wednesday June 25
EIP-8282 go/no-go decision — ACDE Thursday June 18

Detailed notes

Testing Progress

  • •Glamsterdam devnet-5 stable but Prysm has gossip/PeerDAS column issues causing fallback to initial sync; Grandine fixing edge cases00:41:30
  • •No test vectors yet for all remaining devnet-6 EIPs; Spencer targeting Monday for readiness00:44:05

Fork Status And Schedule

  • •EIP-8282 (builder execution requests via system contract) proposed for Glamsterdam; separates builder and validator deposit paths00:07:43
  • •Consensus: EIP-8282 is the right design; correctness over speed; contract audit needed (est. 1-2 months)00:23:52
  • •Splitting EL/CL forks proposed by Dragan; Potuz and Barnabas push back — ePBS needed before gas limit increase due to block propagation constraints00:28:10
  • •Glamsterdam devnet-6 target: next Wednesday (June 25); plan to include EIP-8282 alongside remaining repricing EIPs00:36:06

Eip Proposals Glamsterdam

  • •SSZ Engine API (execution-apis PR #793): 3+ client implementations exist; Prysm nearly ready; not high priority vs. fork work00:37:48

Action items

  • •EL teams review EIP-8282 system contract and confirm implementation scope before ACDE ThursdayEL teams
  • •Review updated execution-apis PR #793 (SSZ Engine API) and align implementations to new specCL + EL teams

Agenda

From the call issue #2116Show

Shared EL-CL Topics

We will have breakouts today...

Glamsterdam

Prep for fast resolution at ACDE #239.

ePBS - EIP-8282 builder deposits (go / no-go?):

  • EIP-8282 #11760 @jtraglia / @wemeetagain / @ensi321 + consensus-specs#5359 - separate builder-deposit path (EIP-7685 request types 0x03 / 0x04), Engine API ExecutionRequests extension, new system contracts (sys-asm#43).
    • Deferred here by ACDC #180 TBD: Include in Gloas? Prelim decision here in ACDT, ratify at the next ACDE.
    • Implementation underway: CL spec #5359, system contracts sys-asm#43, Lodestar PoC lodestar#9507 @markolazic01.
    • vs EIP-8254 deposit cap + O(1) packing - a cap alone is insufficient (the work is in is_pending_validator) per @potuz @tersec @nerolation.
    • Fork-delay risk of a late change. Contract frozen? Who deploys / vanity address (@pk910)?

Other Topics

  • SSZ Engine API (EL-side) @barnabasbusa - status + what is blocking @MariusVanDerWijden's execution-apis#793 (engine Rest-SSZ spec); get client teams implementing. Earlier ask: needs a 3rd CL volunteer.
Devnet Status / Readiness / Scheduling
  • bal-devnet-7 - the last EL-specific devnet; EL benchmarking.
    • New release today: tests-bal@v7.3.2, includes forks <= Prague; no change to Amsterdam coverage.

EL Breakout

Gas Repricings:

  • EIP-2780 rework #11645 @misilva73 - resource-based decomposition of the 21k base; would also resolve/supersede #11735 @rakita.
    • Ratify #11645 as the canonical 2780? Sub-21k transfers now look infeasible - confirm the 21k floor. Two EIP-7702 double-charge edge cases (cold-access, new-account) to close.
  • EIP-8037 (state-create gas, bal-devnet-7 spec merged to EELS #2901):
    • Strict block-gas inclusion check, EIP-literal vs EELS-subtract: land execution-specs#2892 @chfast.
    • EIP-7928 BAL conflict, @rjl493456442 to present the hackmd proposal: charge account-creation in the parent frame + refund - ratify? CALL stays pre-charged (in-frame charging breaks the 2300-gas stipend).
    • Reservoir-based vs source-based refund - decide.
    • EIP-8037 7702 overcharge #11778 @lu-pinto - CHANGES_REQUESTED by @jochem-brouwer, unresolved.
  • EIP-8038 initial numbers #11802 @misilva73 - first concrete state-access numbers (STORAGE_WRITE +257%, STORAGE_CLEAR_REFUND +160%, CREATE_ACCESS +57%).
  • EIP-8038 heads up: Plan is to create a combined eip-8038-2780 branch.
  • EIP-7904 -> Informational #11622 @misilva73 - no compute repricing needed.

Other topics:

CL Breakout

ePBS - fork choice (EIP-7732), @jihoonsong - consensus-specs#5348 Modify get_proposer_head for Gloas

  • get_proposer_head takes/returns ForkChoiceNode; is_parent_strong now counts a PENDING parent's support (#5305) - both Gloas. Also drops is_shuffling_stable (Fulu proposer lookahead; not Gloas-specific).
    • Review + interop implications for the ePBS devnet?

Other topics:

  • QUIC vs mplex (CL networking) @barnabasbusa - aim to decide in the coming weeks whether to drop mplex for beacon p2p. Teku shipped QUIC in its latest stable release; status fro Nimbus? (ACDC #180 had removal at ~Gloas + 2 months; Barnabas wants the decision sooner.)

  • Builder API / buildoor readiness (gates ePBS + 8282 testing): prysm Builder API + auth working @terencechain; (beacon-APIs#588) @nflaig.

  • Staked Builder API builder-specs#138 merged. Heads up.

Last call:

Please comment on this issue with other topics!

Key decisions

  • Correctness over speed: EIP-8282 will be included in Glamsterdam; formal ratification at ACDE Thursday

    IncludedCorrectness over speed prioritized over speed
  • EIP-8282 targeted for glamsterdam-devnet-6 alongside remaining repricing EIPs; devnet-6 target next Wednesday

    glamsterdam-devnet-6

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.