Meeting Summary:
The meeting focused on updates and discussions regarding Blob DevNet Zero, BlockTime 2, and Devnet 3, with emphasis on testing partial cell requests and implementing support for various clients. The team decided to hardcoded cost per state byte for devnet 3’s EIP-8037 implementation and planned to update tests and spec sheets accordingly. They also discussed adding a new testing endpoint in a separate namespace for better benchmarking capabilities, while addressing challenges with ePBS implementation across clients and BlockNet scaling issues.
Click to expand detailed summary
The meeting began with Akash helping with the stream and welcoming participants to ACDT number 71 on February 23rd, 2026. The agenda included updates on Blob DevNet Zero, with Barnabas requested to provide context and updates on this topic. The meeting was still in its opening phase, with participants joining and waiting for the main discussion to begin.
Barnabas reported that Blob.Net Zero is testing partial cell requests and is nearly 80% complete, though some peers experienced disconnections last week that were temporarily resolved by restarting clients. The team is currently implementing support for Lighthouse, Prysm, Teku, Nimbus, and Grandine clients, with Lodestar implementation in progress according to Matthew. The discussion then shifted to BlockTime 2 and Devnet 3, where the main focus was on EIP 8037 implementation, though specific details about progress were not provided in the transcript.
The team discussed implementing EIP-8037 for devnet 3, deciding to hardcode the cost per state byte rather than using the dynamic formula dependent on block gas limit. They agreed to use this approach for devnet 3 and implement the full dynamic version for devnet 4. Spencer committed to delivering the updated tests by ACDE, and Barnabas will update the spec sheet to reference the PR. The team also briefly touched on EIP-7954 and a new testing endpoint for block version number one, with plans to gather client team feedback on making the testing build block version mandatory for devnet 4.
The team discussed adding a new testing endpoint in a separate testing namespace rather than the debug namespace to avoid security issues and enable better benchmarking capabilities. Parithosh explained that the implementation was delayed due to the need for tests, but now that tests are available, the change will be merged this week. Kamil emphasized that this new namespace would not conflict with existing features and would provide significant benefits for test generation and benchmarking. Justin raised concerns about the recurring issue of dependency loops between implementations and tests, highlighting the need for better decoupling.
The team discussed the status of ePBS implementation across different clients, with Stefan reporting that Teku aims to be interop ready by early next week, while Prysm will not meet the current deadline but hopes to reach it the following week. Barnabas proposed setting a realistic launch date for ePBS 0 on March 4th, with a goal of having at least three or ideally four clients ready by then. Kamil provided an update on gas-mit testing, noting progress on repricing efforts but highlighting challenges with long execution times for mainnet-like scenarios, which currently take over a day to complete. The team also discussed issues with BlockNet scaling, including recent outages, and mentioned ongoing work on state load optimization through custom tools and state injection techniques.
The team discussed challenges with the BloTNet test network, particularly focusing on client stability issues at high gas limits and state sizes. Parithosh proposed redistributing keys across multiple clients to better identify where failures occur, rather than having one client handle all block building. The group agreed on the importance of making the bloating process more reliable before increasing scale, with Kamil suggesting parallel efforts to test both current and future state sizes. Additionally, Marcin reported progress on a draft implementation for testing block building with fork support, and Barnabas confirmed successful local interoperability between Lodestar and Lighthouse clients.
Next Steps:
- Barnabas: Investigate the reason for peer disconnects on BlobNet Zero
- Matthew (Lodestar): Complete implementation for BlobNet Zero partial cells by Thursday’s call this week
- Saulius (Grandine): Complete implementation for BlobNet Zero partial cells by next week
- Stefan (Teku): Talk to Lucas about Teku implementation status for BlobNet Zero partial cells
- Spencer: Work on framework changes for EIP 8037 testing until ACDE and provide update at ACDE on Thursday
- Spencer: Add hard-coded cost per state byte change to existing PR for DevNet 3
- Barnabas: Update spec sheet with reference to Spencer’s PR and highlight the edge case for DevNet 3
- Spencer: Release tests by ACDE on Thursday
- Parithosh: Get the testing build block endpoint EIP merged this week
- Justin Traglia: Get reviews from CoreDevs on ePBS-related PRs
- Teku team: Have ePBS interop ready by end of week or early next week
- Prysm team: Aim to have ePBS implementation ready by next week
- Kamil: Check new test inputs from Maria that landed over the weekend
- Parithosh: Answer Will’s questions about Erigon sync on BloatNet
- Parithosh: Take two snapshots of BloatNet - one before Fusaka and one after Fusaka is scheduled
- Parithosh: Try redistributing keys equally and test at 300 million gas to identify where clients break
- Marcin: Continue work on testing build block implementation with fork parameter support for Glamsterdam benchmarking
Recording Access: