EIPsInsight

Loading Experience

Preparing your insights...

Loading0%
All protocol calls
Previous
FCR2026-08-04_012

Fast Confirmation Rule #012

2026-08-04 3 decisions 535 transcript lines

Transcript

Call summary

Decisions
  • •FCR recurring breakout call series concluded; future coordination via ad hoc calls and Telegram - 00:06:14
  • •No change made to safeBlockHash fallback behavior (FCR disabled → justified); deferred pending clarity on who relies on justified block hash - 00:27:12
  • •PTC-based improvement to FCR for post-Gloas slot structure not pursued now; decoupled Consensus will supersede current FCR - 00:53:36
Highlights
  • Client Updates:
    • ·Grandine: PR ready, 3 failing tests due to known spec issue; awaiting fixed test vectors - 00:07:25
    • ·Lodestar: FCR implementation working on mainnet; a few spec tests disabled in CI pending updated vectors - 00:09:03
    • ·Prysm: FCR branch passing latest nightly spec tests; not yet run on mainnet or devnet; deeper review still needed - 00:11:00
    • ·Teku: Merging FCR feature branch to master in smaller chunks; ~2–3 weeks remaining; investigating Lighthouse comparison failure - 00:12:10
  • Fcr Post Gloas:
    • ·Post-Gloas: FCR confirms payload from 2 slots ago due to new slot structure; using PTC votes to recover the previous slot's payload is under consideration but deemed complex and risky - 00:46:25
    • ·Decoupled Consensus will ship its own confirmation rule, likely deprecating current FCR; improvement work may not be warranted given this roadmap - 00:53:36
  • Organizational:
    • ·This is the last recurring FCR breakout call; future calls will be ad hoc as needed - 00:06:14
  • Testing Progress:
    • ·Glamsterdam devnet FCR report published by EthPandaOps; waiting on Prysm and Teku for full client comparison run - 00:14:22
    • ·Client comparison on historical data via FCR simulator deferred until Prysm and Teku are production-ready - 00:16:01
    • ·Several consensus-specs PRs merged (#5449, #5489, #5498, #5499): fix safe execution block hash bug and disabled-BLS test issues - 00:16:46
  • Fcr Safe Block Hash:
    • ·Proposal: set safeBlockHash to finalizedBlockHash (not justified) when FCR is disabled; safety of justified is questionable - 00:21:20
    • ·Potuz: Arbitrum was using SAFE for deposits; likely runs own nodes so probably unaffected by this change - 00:29:56
    • ·Only Nimbus enables FCR by default; all other clients use a feature flag, so operators opting in understand the implications - 00:31:16
  • Fcr Node Restart Behavior:
    • ·On CL restart, safeBlockHash may reset to finalized; behavior is implementation-dependent (Lighthouse persists fork choice to disk; Prysm does not) - 00:32:16
    • ·Consensus: caching the confirmed root across restarts is safe provided it is not older than 2 epochs; EL could perform this fallback itself - 00:41:25
Action Items
  • •Mikhail Kalinin / Justin Traglia - Post thoughts on fuzzing FCR implementations to the Telegram group; contact Nikos (EF security team) re: agent-swarm review - 00:18:42
  • •Mikhail Kalinin - Once Prysm and Teku FCR implementations are production-ready, ask EthPandaOps to re-run cross-client historical comparison simulations - 00:16:01

Key decisions

  • FCR recurring breakout call series concluded; future coordination via ad hoc calls and Telegram

    Future coordination will be via ad hoc calls and Telegram

  • No change made to safeBlockHash fallback behavior (FCR disabled → justified); deferred pending clarity on who relies on justified block hash

    Deferred pending clarity on who relies on justified block hash

  • PTC-based improvement to FCR for post-Gloas slot structure not pursued now; decoupled Consensus will supersede current FCR

    Decoupled Consensus will supersede current FCR

Links