ACDT2026-11-16_103

AllCoreDevs - Testing #103

2026-11-16 3 decisions 738 transcript lines

Transcript

Call summary

Targets
  • •Glamsterdam mainnet targeting January 2027 (per protocol signals from affected teams) - 01:33:29
  • •EIPIP call: Wednesday November 18, 16:00 UTC — AI integration into EIP review process - 01:51:19
Decisions
  • •No decision on EIP-8037 today; formal path-forward decision deferred to ACDE on Thursday - 01:30:30
  • •Glamsterdam delay was justified to prioritize app-layer mitigation time before mainnet - 01:17:50
Highlights
  • Eip Process:
    • ·jochem-brouwer: large influx of AI-generated EIPs; EIPIP call Wednesday Nov 18 16:00 UTC to discuss AI integration into review process - 01:50:21
  • Audit Findings:
    • ·ChainSecurity: 152 findings across mainnet; reservoir/refill mechanism is minority of breakage - 00:28:45
    • ·Dedaub: 12 high + 70 medium confirmed findings from ~25K contracts with non-trivial gas patterns - 00:37:30
    • ·EF internal: 3 highs (1 fixed, 2 fixing within a month); at least 25 confirmed mediums; 16 actively triaging - 00:46:52
    • ·OP Stack chains independently affected; OP has fixed invariants, live on OP Sepolia - 00:48:25
  • Process Improvements:
    • ·Toni: core devs must clearly define which invariants are guaranteed vs. likely to change - 00:59:00
    • ·Marius: proposal to publish EIP documenting stable vs. expected-to-change protocol invariants - 01:20:21
    • ·Specs team (Mario/danceratopz): two open PRs adding deployed-idiom invariant corpus and chain invariant checks - 01:22:57
    • ·EF/Ethlabs agent-swarm analysis was largely automatable; will be run systematically for future EIPs - 01:39:25
    • ·EF Access Layer (Rahul): builder enablement team surveyed 50+ teams; direct contacts for 3,000+ L1 devs - 01:16:18
  • App Developer Concerns:
    • ·Jango (Juicebox): EIP-8037 allows users to tune gas and trigger catch blocks, breaking composability - 00:50:07
    • ·Daniel (Uniswap): reservoir mechanism introduces implicit user-controlled branching; fundamentally breaks try-catch - 01:01:38
    • ·Neville (Dedaub): no reliable way for callers to distinguish OOG revert from logical revert - 00:59:00
    • ·Barnabas: Ethereum cannot honestly promise true immutability given known upcoming PQ/state changes - 01:14:14
  • Glamsterdam Eip8037 Impact:
    • ·EIP-8037 and EIP-8038 break smart contract invariants; central topic of call - 00:17:49
    • ·EIPs enable 3.3x gas scaling (60M → 200M gas/block) but break hard-coded gas patterns - 00:21:37
    • ·Delays in audits and outreach caused by EF reorgs and budget cuts; Marius accepts responsibility - 00:21:43
    • ·Combined search: 200 total findings (152 ChainSecurity + EF internal); coverage not complete - 00:30:31
    • ·Majority of findings from Invariants 4 and 7: hard-coded gas budgets insufficient post-Glamsterdam - 00:31:04
  • Protocol Rationale And Tradeoffs:
    • ·~10–15K contracts affected out of ~90M on Ethereum; immutable broken contracts hold ~$300–500K - 00:54:33
    • ·No silver bullet: any 2D gas version breaks something; removing 8037 also blocks large contract deployment - 01:39:54
    • ·Affected protocols signaling willingness to fix indicate a January Glamsterdam date is acceptable - 01:33:29
    • ·Glamsterdam is a precondition for Hegota (PQ); removing it delays post-quantum accounts by ~4 months - 01:32:09
Action Items
  • •EF specs team - Review and merge specs PRs #3948 and #3982 (deployed-idiom invariant corpus and chain invariant checks) - 01:22:57
  • •All core devs (ACDE Thursday) - Decide on Glamsterdam path forward (ship 8037 as-is vs. changes) and agree on Hoodi date - 01:30:00
  • •Rahul (EF Access Layer) - Coordinate outreach to remaining affected L2/app teams not yet triaging; use direct security contacts - 01:17:50

Key decisions

  • No decision on EIP-8037 today; formal path-forward decision deferred to ACDE on Thursday

    The decision on whether to ship EIP-8037 as-is or implement changes was deferred to the upcoming ACDE meeting.
  • Glamsterdam delay was justified to prioritize app-layer mitigation time before mainnet

    The delay in the Glamsterdam upgrade was deemed necessary to allow application developers sufficient time to address breaking changes caused by new EIPs.
  • Glamsterdam mainnet targeting January 2027

    SFIThe upgrade is targeted for January 2027 based on signals from affected protocols indicating readiness to implement fixes.

EIPs discussed

AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.