All Core Devs - Testing (ACDT) #41 | June 23 2025

[marioevz]

All Core Devs - Testing (ACDT) #41, June 23 2025

• June 23, 2025, 14:00 UTC

Agenda

• Fusaka Devnet 2
• PeerDAS testing
• EIP-7907 and tests
• engine_getBlobsV2: engine_getBlobsV2: add partialResponse flag to enable partial hits execution-apis#669

Other comments and resources

The zoom link will be sent to the facilitator via email
Facilitator emails: XXXXX, YYYYY

🤖 config

• Duration in minutes : 60
• Recurring meeting : true
• Call series : ACDT
• Occurrence rate : weekly
• Already a Zoom meeting ID : true
• Already on Ethereum Calendar : true
• Need YouTube stream links : true # Set to false if you don't want YouTube stream links created
• display zoom link in invite : false # Set to true to add the Zoom link to the Google Calendar invite description

[github-actions]

Note: Zoom meeting creation skipped as requested in issue.

Note: Google Calendar event creation skipped as requested in issue.
Discourse Topic ID: 24594

• Action: Updated at 2025-06-23 12:45 UTC
• URL: https://ethereum-magicians.org/t/24594

Existing YouTube Stream Links (Reused):

• Stream 1 (Jun 23, 2025): https://youtube.com/watch?v=3mnaILTNk6w
• Skipping email notification (already processed issue All Core Devs - Testing (ACDT) #41 | June 23 2025 #1583 before).

Telegram Notification

• Sent message for this occurrence to Ethereum Protocol Updates channel

[jochem-brouwer]

I want to bring up EIP-7907 in light of benchmarks I've done on a mainnet shadowfork performance network, the current and proposed state of the EIP, the recent discussions and my thoughts on it (we should ship a code size bump! 😄 👍 )

[rakita]

About fusaka EIP-7907 devnet2 tests : https://github.com/ethereum/execution-spec-tests/releases/tag/fusaka-devnet-2%40v1.2.0

Have noticed this behaviour on eip7907_meter_contract_code_size/meter_contract_code_size/contract_reentry.json test

Test is about `to` account with big contract getting called again with CALL opcode.

I had bug here where to gets loaded warm but gas is not spend, and on CALL this account code would be warm.

Would assume that we should we deduct on start (when loading to) but I can't deduce this 100% from test.

Nice test btw
draganrakita — 11:52
Looking at other tests that I am passing, they don't deduct this initial to gas, and on second thought this would introduce even more edge cases. 

So would say that test assumes initial `to` to stay cold. This feels like a bug

Thread from discord: https://discord.com/channels/595666850260713488/1386500548525559909/1386645006475067545

[raulk]

I'd like to discuss this small tweak to engine_getBlobsV2: ethereum/execution-apis#669

[marioevz]

Added all items to agenda 👍

[SunnysidedJ]

Sorry for late update, could we (Sunnyside Labs) shortly present the devnet results from Berlin interop?