Agenda
- blob-devnet-0
- bal devnets or glamsterdam devnet
- epbs-devnet-1
- engine: EL must support reorg to head’s ancestor
- debug RPC update
Meeting Time: Monday, April 20, 2026 at 14:00 UTC (60 minutes)
Meeting Time: Monday, April 20, 2026 at 14:00 UTC (60 minutes)
The meeting focused on discussing DevNets, with particular emphasis on Block-level Access List DevNet 4 and related specifications. The team debated whether to proceed with BalDevNet 4 or switch to BalDevNet 3 for Glamsterdam DevNet Zero, ultimately deciding to move forward with BalDevNet 4 implementation. They also reviewed a PR regarding execution APIs and block reorgs, discussing the proposed 32-block limitation and its implications for EL-CL synchronization. The conversation ended with a brief discussion about potentially removing payload envelope syncing on the CL side, though this topic was tabled for further async discussion. Breakout rooms were established to continue discussions on specific topics, including EPBS and block-level access list implementation details.
The team discussed DevNets, with Barnabas reporting that most CL teams are focusing on shipping EPBS, with only Prism and Lighthouse currently deployed. Barnabas mentioned enabling partial cells on every node and suggested starting onboarding other clients. Nico indicated that their team still has some work to complete this week. Potuz mentioned they would like to discuss syncing both here and ACD eventually.
The team discussed moving forward with Glamsterdam DevNet development, with Marius initially proposing to either use ValDevNet4 or Romsterdam DevNet Zero. After discussion, Barnabas and Potuz advocated for using both networks simultaneously, with Glamsterdam DevNet Zero based on BAL DevNet 3 while continuing development on BAL DevNet 4 to implement EIP-8037 changes. The team agreed to launch both networks, with Marius confirming that the EIP-8037 implementation would require a rebase due to recent master branch changes but not because of the dynamic values themselves.
The team discussed transitioning to BalDevNet 4 instead of BalDevNet 3, with Maria and Marius advocating for this change due to implementation differences and the complexity of dynamic cost testing. While Barnabas raised concerns about potential bugs in BalDevNet 3 affecting context switching, the team agreed that using BalDevNet 4 would allow for better testing alignment and meet Parithosh’s timeline of late this week or early next week for demos. Marius estimated it would take 1-2 days to complete the necessary changes and rebase the work, while Spencer noted that additional testing would be required for the 7 spec changes to 8037.
The team discussed the readiness of EL teams for 10.4 changes, with Barnabas seeking confirmation about additional clients being ready by Wednesday. Spencer confirmed tests would be completed by end of day or early the next day, while Marius suggested undoing the dynamic state cost change as an easier testing approach. The discussion included clarification about the DevNet 4 specification, though there was some confusion about whether variable state cost removal was planned, with Stefan expressing concern about this change. Barnabas noted that while some teams were previously uncertain about meeting the end-of-week deadline, he was still waiting for more client confirmations.
The team discussed a PR related to execution layer (EL) behavior when handling reorgs and payload validation. They debated whether EL clients should be required to support reorgs up to the finalized block or implement a specific limit, ultimately agreeing that a 32-block limit was reasonable. The group also touched on payload envelope syncing optimization, though this discussion was not fully resolved and will be continued asynchronously. The conversation ended with plans to move to breakout rooms for further discussions on specific topics.
DT!l3mN3)DT!l3mN3)DT!l3mN3)YouTube Stream Links: