All Core Devs - Execution (ACDE) #212 (May 22, 2025)

[timbeiko]

• May 22, 2025, 14:00-15:30 UTC
• Stream
• Ethereum Protocol Calls Calendar subscription

Agenda

• Fusaka planning
  • SFI EIPs for devnet-1
  • DFI EIPs that won't make it in devnet-2
• Testnet deployment strategy
• Glamsterdam planning process
  • Community Consensus, Fork Headliners & ACD Working Groups
  • Glamsterdam scoping proposal
• ACD Breakout Room Reviews
• ACD input collection
  - https://ethereum-magicians.org/t/for-an-acd-platform/24098
  - https://ethereum-magicians.org/t/all-core-devs-acd/24198
• Review requests
  • Standardize JSON-RPC Error codes execution-apis#658
  • https://eips.ethereum.org/EIPS/eip-7939
  • Add EIP: Indexed Storage EIPs#9792

🤖 config

• Duration in minutes : 90
• Recurring meeting : true
• Call series : ACDE
• Occurrence rate : bi-weekly
• Already a Zoom meeting ID : false #
• Already on Ethereum Calendar : false #
• Need YouTube stream links : true #
• Facilitator email: tim@ethereum.org
  Note: The zoom link will be sent to the facilitator via email

[github-actions]

Discourse Topic ID: 24118

• Action: Updated at 2025-05-22 14:03 UTC
• URL: https://ethereum-magicians.org/t/24118

Zoom Meeting: Reusing meeting 88269836469 for series 'acde'.

• Zoom link hidden (sent to facilitator email)

Existing YouTube Stream Links (Reused):

• Stream 1 (May 22, 2025): https://youtube.com/watch?v=FEZGRUPkI_8

Calendar Event Updated

• Google Calendar
• Skipping email notification (already processed issue All Core Devs - Execution (ACDE) #212 (May 22, 2025) #1533 before).

Telegram Notification

• Updated message for this occurrence to Ethereum Protocol Updates channel

[benaadams]

I'd like to consider adding CLZ to Fusaka

https://eips.ethereum.org/EIPS/eip-7939

CLZ isn't CFI'd but if we want to send signal we are listening to dev community, would be a good one? (Is fairly a simple but high impact OPCODE)

[alonmuroch]

I'd like to propose a discussion (and present an idea) around improving the stability of our testnets to provide a smoother experience for application developers.

In the spirit of my post on Ethereum Magicians, I'm suggesting the following for consideration and feedback:

• Establish Hoodi as a long-standing testnet dedicated to application development.
• Exclude Hoodi from protocol fork testing; instead, it should be forked from mainnet as closely as possible to avoid disruptions like those we’ve previously encountered.
• Encourage developers to maintain a “staging” deployment of their contracts on Hoodi, providing integrators and testers with a rich, stable environment.
• Ensure that protocol forks are tested elsewhere first, before being applied to Hoodi, to minimize potential issues as much as possible.

[benaadams]

Could we discuss SFI'ing for next devnet

EIP-7825: Transaction Gas Limit Cap

As it resolves an issues with increasing gaslimit (how does a user estimate gas on a tx that is too large in bytes, parallelization and fairness)

Also discuss the appropriate values for this EIP

e.g. for Tx Cap is 5M 10M 15M gas better than the 30M gas proposed

Considering all recent txs > 5M are decomposable (and 90% are XEN mining); a higher tx cap makes parallelizing harder; as if a block is a single tx consuming all gas it can only be run serially; as well as fairness for allowing other txs.

[benaadams]

Could we discuss SFI'ing for next devnet

EIP-7934: RLP Execution Block Size Limit

As it resolves the issue with block size breaking the network layer so preventing impossible to propagate blocks from being valid

Also discuss whether 10MiB - 512KiB is the right size or if it needs to be more like 7.5MB to allow for 2.5MiB of slashings etc at the CL

Additionally whether this size is still too large for timely propagation on p2p given the upload recommendation of 50MBit/s; how many peers must it be sent to initially etc (10MiB being 80MBits)

[abcoathup]

Re Marc's point about a new platform for ACD.
I have created a pinned wiki topic on Eth Magicians as a portal for ACD info.
https://ethereum-magicians.org/t/all-core-devs-acd/24198

We can experiment with the type of information we want to surface using the existing Eth Magicians platform.

[simsonraj]

Hello all,
I'd appreciate an opportunity to discuss our proposal to standardise JSON-RPC error codes across different execution clients
ethereum/execution-apis#658

[timbeiko]

@benaadams we're well past the deadline for CFI'ing EIPs for Fusaka so unless there's an overwhelming amount of community support, full test coverage, and strong client consensus for EIP-7939, then I don't think we should re-open the Fusaka CFI conversation. That said, thank you for sharing your SFI preferences, we'll hopefully finalize those on the call.

@alonmuroch queued you up for right after the Fusaka planning conversation 😄

@abcoathup thanks for putting this together! I think we should touch on it as part of the. Glamsterdam planning convo if time permits. As a side note, I wonder if we should improve the /pm repo to be clearer about all this info.

@simsonraj added it to the agenda if time permits at the end. If you have specific questions or input you'd like from client teams, I'd suggest giving more context in your comment.

[simsonraj]

Thank you @timbeiko. Appreciate it
This is mainly to let the client teams know this effort is underway & gather first round of feedback on the proposal before implementing it fully, as we have the framework ready to implement the solution.
The proposal PDF is attached to the issue.
Wanted to understand if the range of error codes defined in the proposal is good & the error groupings make sense for the users & clients alike, or a better grouping, if any.

[keyvank]

Hey @timbeiko
May I also present this EIP on the call? 😃 ethereum/EIPs#9792

[abcoathup]

improve the /pm repo to be clearer

@timbeiko /pm repo needs an overhaul.

For each call we should have a file stored with:

• Agenda (copied from the Issue)
• Moderator summary of actions/decisions (copied from Eth Magicians, otherwise we change to having in the Issue)
• Raw transcript
• AI & (optionally) human improved transcript
• AI generated summary
• (Ideally) audio recording
• Raw chatlog
• Threaded chatlog (need to convert raw)
• Each calls files should be in a folder, matching the current call naming

Needs to be automated so we only need a human to review human improvements to transcripts.
@nconsigny has been doing a great job on automation, so ideally this would be a continuation of that.

[jflo]

Besu Fusaka Scheduling updated opinions:

devnet - 0
EIP-7594 : PeerDAS, cell proofs on tx receipt
EIP-7892: Blob Parameter Only Hardforks
EIP-7823: Set upper bounds for MODEXP
EIP-7883: ModExp Gas Cost Increase

devnet-1
EIP-7825: Transaction Gas Limit Cap
EIP-5920: PAY opcode
EIP-7934 : 10MB block limit enforced on block production
^^^^^^^^ - can we discuss why this limit should be in consensus, as opposed to an agreed operational limit?

devnet-2
RIP-7212: Precompile for secp256r1 Curve Support
EIP-7907: Meter Contract Code Size And Increase Limit
EIP-7918: Blob base fee bounded by execution cost

Remove from CFI:
EIP-7762: Increase MIN_BASE_FEE_PER_BLOB_GAS
^^^^^^^^ - can an economist explain to us why we'd want this over the more dynamic 7918?