All Core Devs - Testing (ACDT) #77, April 13, 2026

Protocol Calls & happenings
,
1

Agenda

Agenda

  • blob-devnet-0
  • bal devnets
  • epbs-devnet-1
  • debug RPC update

Meeting Time: Monday, April 13, 2026 at 14:00 UTC (60 minutes)

GitHub Issue

1 Like
2

Meeting Summary:

The DevNet team met to discuss updates and planning for upcoming network releases. Barnabas reported that EPB DevNet has been broken since launch and progress is being made on simplifying beacon state for a relaunch of DevNet 2. Stefan updated on the launch of DevNet 3, which encountered issues with Geth clients due to gas handling discrepancies, particularly around state gas versus regular gas usage. The team discussed EIP 8037 changes and whether to include them in DevNet 3 or 4, with Spencer and Maria noting alignment between current clients and spec but questioning the current design. The group decided to include EIPs 7976 and 7981 in DevNet 4, aiming for launch before the interop event. A significant discussion arose around a proposed change to EPBS to enable range syncing without requesting payloads, with concerns raised about potentially syncing through invalid blocks. The conversation ended with plans to move detailed discussions to breakout rooms.

Click to expand detailed summary

The team discussed updates on DevNet, with Barnabas reporting progress on APPS development and plans for a breakout session to decide on changing the beacon state for DevNet 2 relaunch. Stefan provided an update on BoudafNet 3, launched on Wednesday, which encountered issues with Geth clients due to discrepancies in gas handling between clients. The team identified a need for clarification on EIP behavior and discussed potential fixes for BoudafNet 3 and future versions.

The team discussed the 8037 topics and agreed to address them in breakout rooms. Stefan mentioned creating a spec sheet for DevNet 4 and suggested adding 2780 to it, as it affects every transaction. The group reviewed the major changes between DevNet 3 and 4, with Spencer explaining that most of DevNet 3’s current specifications have open questions. Spencer clarified that for DevNet 3 to 6, they need to add spec clarifications to the EIP text and include tests, while DevNet 4 will focus on modifications that change the spec and tests, including adding dynamic cost per state bike for EIP 8037.

The team discussed potential changes to include in DevNet 4, with Maria suggesting a discussion on points 3 to 6 in her document and noting concerns about the current design. Stefan proposed adding items 7976 and 7981 to the discussion, while Ben suggested including 2780 due to its broad impact on transactions. The team debated implementation complexity and timeline, with Stefan aiming for a launch before interop by April 29th, though Ben and others noted that 2780 would require significant updates to tests. Danceratopz confirmed that a PR for execution speculation for 2780 was already available with new tests implemented.

Maria explained that EIP 2780 can begin implementation before final numbers are available for state access EIP 8038, as most work involves breaking down transaction costs into components, with parameter values to be updated later. Toni agreed and highlighted concerns about the low cost for calling EOA accounts, which could introduce a new worst-case block size, suggesting this needs reconsideration. Both participants noted that the data EIPs are small and contained, and while there is work to be done with EIP 8037, they can potentially proceed with EIP 2780 implementation.

The team discussed including EIPs 7976 and 7981 in BalDevNet 4, with Toni confirming they should be included based on high client priority rankings and minimal ecosystem concerns. The group agreed to aim for implementation before the interop event, with Barnabas expressing preference for stability over additional EIPs. The team decided to defer further discussion of EIPs to a breakout room session, with Parithosh noting there was more relevant material to cover in the breakout discussions.

Potuz discussed a proposal to change PBS to allow range sync without requesting payloads, which would allow the consensus layer to handle all state transitions without validating payloads. He raised a concern about whether this could allow clients to sync through finalized invalid blocks, particularly for unfinalized sections of the chain. Parithosh suggested addressing this topic immediately as a higher priority than AD37 changes.

The team discussed the handling of invalid blocks during sync processes, particularly in scenarios involving snap syncing and backward syncing. Łukasz explained that Nethermind validates blocks during replay and would catch invalid blocks, but Potuz raised concerns about not checking block hashes without accessing payloads. The group agreed that while there are potential risks of getting stuck if invalid blocks cannot be obtained from the network, the current approach should be acceptable, with Nethermind confirming they can handle this scenario. Potuz requested validation from other EL developers before implementing this approach.

The team discussed documenting certain information in writing, with Potuz agreeing to provide a quick writeup. They addressed issues with breakout room functionality, where several participants, including Potuz, nixo, and Eitan, were unable to see or access the breakout rooms. Parithosh worked to resolve the technical issues and provided instructions for participants to access breakout rooms through the menu bar.

Recording Access:

3

YouTube Stream Links:

4

Video, transcript & chatlog

News coverage

Resources