All Core Devs - Testing (ACDT) #71, February 23, 2026

[marioevz]

UTC Date & Time

February 23, 2026, 14:00 UTC

Agenda

Fusaka:

• blob-devnet-0 updates
• partial cells / getBlobv3 implementation update

Glamsterdam:

• bal-devnet-2 updates
• bal-devnet-3 readiness
  • EIP-8037 specs/tests status
• epbs-devnet-0 implementation update
  • client readiness check
• Adding testing_buildBlockV1 for future devnets (4+)
  • Client teams readiness

Gas limit:

• TBD

State bloat:

• perf-devnet-2 updates

Call Series

All Core Devs - Testing

Autopilot Mode

• Use autopilot (recommended defaults for this call series)

🔧 Meeting Configuration

Duration

60 minutes

Occurrence Rate

weekly

Use Custom Meeting Link (Optional)

• I will provide my own meeting link

Display Zoom Link in Calendar Invite (Optional)

• Display Zoom link in invite

YouTube Livestream Link (Optional)

• Create YouTube livestream link

[github-actions]

⚡ Protocol Call Resources:

✅ Zoom: Join Meeting
✅ Calendar: Add to Calendar
✅ Discourse: Discussion Topic
✅ YouTube Live: Watch Live

[parithosh]

I'd like to bring up the inclusion of testing_buildBlockV1 (ethereum/execution-apis#747) for bal-devnet-4 onwards (so not the next devnet, but the one after). cc @MysticRyuujin

[marioevz]

@yperbasis via discord on Erigon status for devnet-2:

We've finished our bal-devnet-2 implementation (kudos to Mark Holt) and pass all the tests more or less. We're working on a couple of remaining flaky tests and pesky races in parallelized execution.

[spencer-tb]

I want to briefly bring up some issues we've been having with the EELS testing framework for 8037. The EIP adds a new paradigm at a high level on the framework: gas costs are now dependent on the block gas limit via cost per state byte. This impacts not only the framework but a lot of tests >600 files. We're working on a good framework solution for this but if we don't get there by ACDE we might want to consider making the cost_per_state_byte a fork constant, this massively reduces he complexity on the framework side, and will enable us to get tests out.

Similary on 8037, how do we want to tackle tracing (originally brought up by SamWilsn). We have added stateGas and stateGasCost to the EELS tracer alongside gas and gasCost. Should we update EIP-3155 with this?

[MariusVanDerWijden]

Making cost_per_state_byte a fork constant would kinda defeat the purpose of 8037 tbh.
The whole idea is to be able to change the gas limit independent of hardforks, otherwise we could just go with fixed costs. Maybe a way forward would be to postpone 8037 to devnet-4 to give testing teams more time to update their framework with this new paradigm. But I do think we will use this paradigm more and more in the future, so it makes sense to spend some time to create a good solution for it

I think we should update 3155 with the stateGas and stateGasCost.

[akashkshirsagar31]

X Stream: https://x.com/i/broadcasts/1yxBeMMWnenJN

[jtraglia]

For consensus specifications, I would like to mention:

• EIP-7805 (FOCIL) specifications have been promoted to Heze specifications.

Request core devs review the following PRs related to Gloas:

• Variable PTC deadline consensus-specs#4843
• Remove impossible branch in forkchoice consensus-specs#4892
• Remove pending status from tiebreaker consensus-specs#4898
• Only allow attestations for known payload statuses consensus-specs#4918
• Request missing payload envelopes for index-1 attestation consensus-specs#4939

Request core devs review the following PRs unrelated to Gloas:

• Add helpers for simple config derivations consensus-specs#4814
• Add executable gossip validation functions for phase0 consensus-specs#4902
• Replace SECONDS_PER_SLOT with SLOT_DURATION_MS in specs/tests consensus-specs#4926

PS: I do not intend to give an overview of these PRs during ACDT.

[bomanaps]

Just for more spotlight on this https://hackmd.io/@bomanaps/rJiHuPFOZl

[github-actions]

This meeting occurred more than 24 hours ago. Closing automatically.