ACDC #185 · EIP-7610 Excluded from Glamsterdam
AllCoreDevs - Consensus #185 · 2026-08-20
Drops an EIP that only added performance burden, and EIP-8253 solves the problem instead.
EIP-7610 was officially cancelled in Glamsterdam, and four EIPs, including EIP-8253 which replaces it, were accepted as candidates for the next fork, Hegotá. Testnet fork dates were also proposed for the first time: Sepolia on September 28 and Hoodi on October 26.
Why it matters
With testnet fork dates proposed, the Glamsterdam mainnet rollout is coming into view. If Sepolia, the dapp-focused testnet, forks first, application developers can test the gas repricing changes ahead of time.
Decisions
- EIP-7610 excluded: EIP-7610 was cancelled outright in Glamsterdam rather than deferred. EIP-8253 in Hegotá removes the root cause, so this rule itself is no longer needed.
- Non-finality devnet: It was decided to launch a non-finality devnet the week after next. Before that, every client must land its code on the trunk branch.
- EIP-8333 PFI: EIP-8333 was accepted as Proposed for Inclusion (PFI) in Hegotá. This means a proposal that fixes an off-by-one error in checkpoint calculation has become an official candidate.
- EIP-8379 PFI: EIP-8379 was accepted as Proposed for Inclusion in Hegotá. This means an Engine API extension that makes the CL the single source of truth for sync has become a candidate.
- EIP-8383 PFI: EIP-8383 was accepted as Proposed for Inclusion in Hegotá. This means the proposal to halve the CL block retention period stays on as a candidate.
- EIP-8253 PFI: EIP-8253, which replaces EIP-7610, was accepted as Proposed for Inclusion in Hegotá. It makes the problematic account type impossible to exist in the state trie at all.
Discussion
Can the builder deposit cache be trusted across a fork transition
Six hours before the call, when the Gloas fork activated on devnet 8, several clients stalled on cache problems while processing about 16,000 builder deposits, and a debate began over whether the same design is safe on mainnet.
EIP-8379: can the Engine API be optional
The Nimbus side proposed an Engine API extension that makes the CL the single source of truth for sync as a Hegotá candidate, and the question became whether it can be an optional implementation like other networking EIPs.
Outcome: PFI decision 00:54:23
Does the CL block retention period need to be reduced
A proposal (EIP-8383) emerged saying that the constant underlying the retention period should be halved because of the introduction of EIP-8061 in Glamsterdam, but opinions split from the start on why it should change at all.
Outcome: PFI decision 00:54:23
Hegotá tier rankings: the same A means different things
All client teams agreed to publish their Hegotá candidate EIPs sorted into S/A/B/D tiers by mid-September, and how to read the tier labels immediately became a point of debate.
Glossary
- SFI: Scheduled for Inclusion, the state where inclusion in the fork is confirmed
- CFI: Considered for Inclusion, the state where the EIP is being reviewed on the premise that it will be included
- PFI: Proposed for Inclusion, a status in which inclusion has been proposed and the proposal has been accepted as a candidate
- DFI: Declined for Inclusion, the state of being removed from candidacy
- ACDC: The biweekly Ethereum consensus layer (CL) core developer call
- Non-finality: A state in which the chain fails to reach finality. It is triggered deliberately to test clients
- DevNet: A temporary network where developers test new code among themselves before it reaches a testnet
- Engine API: The internal API through which consensus layer (CL) and execution layer (EL) clients communicate with each other
EIPs covered
- EIP-7610 Reject contract creation at addresses with leftover storage
- EIP-8061 Higher exit and consolidation processing limit
- EIP-8253 Raising the nonce of nonce-0 storage accounts
- EIP-8333 Aligning checkpoints to the epoch boundary block
- EIP-8379 Top-up sync: consensus-layer-led execution-layer sync
- EIP-8383 Shorter consensus layer block retention period