Reporting window: October 12, 2026, 00:00:00 to October 19, 2026, 00:00:00 WITA (Asia/Makassar), end exclusive.

Glamsterdam’s rollout reached a consequential boundary: developers chose to preserve tested fork behavior despite an application-layer objection. The coordination repository merged Hoodi’s schedule, while AllCoreDevs Consensus declined a late builder-onboarding change requested by Lido. Hegotá’s scoping discussion also declined Optional Execution Proofs and RowDAS. The editorial implication is practical: deployment preparation now deserves more attention than expanding either upgrade’s ambitions. The strongest new testing evidence concerns assumptions about deposit routing, inclusion-list timing, and execution order.

Hoodi’s schedule leaves staking applications with a migration problem

Nixo’s merged scheduling change sets Hoodi at epoch 132352, October 27 at 01:42:48 WITA, or October 26 at 17:42:48 UTC. The same change leaves mainnet timing as TBD. On the call, Alex Stokes identified Lodestar and ethrex releases and requested the remaining releases. That is a dated readiness report, not confirmation that every client met the deadline.

The harder issue came from Lido’s Dmitrii Gusakov. The reported fork-onboarding behavior can credit a queued validator deposit to a builder with the same public key. A staking application that expects a validator record can consequently lose the evidence needed for its normal recovery path. The issue describes that consequence for protocols whose checks depend on the validator appearing; it does not establish that every staking protocol is affected.

Gusakov requested a routing change. Stokes closed the discussion without changing the fork behavior, after developers raised release timing and retesting concerns. The immediate watchpoint is therefore application-side preparation and builder-onboarding tests. Mainnet timing remains contested: Barnabas Busa proposed December 1, with December 8 as an alternative; Gusakov preferred December 8 for more testing with real builders. Neither comment announces an activation date.

Hegotá’s scoping separates experimental work from fork commitments

Stokes called DFI for EIP-8025: Optional Execution Proofs after finding insufficient consensus among consensus-client teams. Potuz’s objection concerned the maintenance obligations that optional functionality would impose on subsequent specification work. EIP-8371: RowDAS — Distributed Blob Reconstruction also received DFI, with client support and current blob demand central to the discussion.

Those are fork-scoping outcomes, not findings that either research direction lacks merit. By contrast, EIP-8205: Withdrawal credentials preregistration remained deferred to the next call, following Greg K’s request for more discussion. Watch whether the next call resolves that proposal and whether execution-proof contributors obtain a workable integration process outside Hegotá’s scope.

Inclusion-list timing gives committee members another second

Jihoon Song’s merged consensus-specification change moves the inclusion-list deadline from eight to nine seconds into the slot. Song argues that processing a late payload can leave committee members too little time to construct and broadcast a list. The change gives builders one second less to incorporate lists; Song explicitly calls the deadline tentative pending mainnet data. The merge establishes a specification parameter, not a demonstrated timing margin across clients.

Execution testing gained a complementary check: spencer-tb’s merged simulator change checks inclusionListSatisfied and probes engine_getInclusionListV1 after importing inclusion-list fixtures. The useful next signal is a client-by-client result under the revised timing and API checks. Neither merge establishes that every implementation passes.

Execution tests expose assumptions that ordinary contracts conceal

fselmo’s merged withdrawal-order tests use synthetic request contracts to make processing order observable. The author reports Reth state-root mismatches and Geth Amsterdam BAL-hash mismatches in the forward_all cases. Besu, Nethermind, Erigon and ethrex passed the reported tests. The same report says deployed request contracts do not read balances or send ether, so the ordering difference remains latent on real chains. The finding warrants follow-up; it does not demonstrate an ongoing production split.

Two other merged tests sharpen gas-accounting boundaries. Busa’s floor-inclusive gas test distinguishes remaining block gas from remaining execution gas for Prague through Osaka. fselmo’s canonical-order tests reject allowances that would fit only after reordering transactions, a relevant check for parallel BAL execution. Watch fixes and reruns against named client versions before treating the new coverage as resolved interoperability.

Sources and further reading