Reporting window: September 14, 2026, 00:00:00 to September 21, 2026, 00:00:00 WITA (Asia/Makassar), end exclusive. This corrected edition covers the complete ISO week 38.

Glamsterdam gained a firmer public-testnet schedule this week: Alex Stokes’s Sepolia activation update entered the merged upgrade record. At ACDC #187, pk910 also reported a smooth devnet-11 transition with no participation drop. The editorial assessment is qualified progress. A transition report strengthens the release case, but client behavior under sustained stress and the reliability of test fixtures still deserve attention. Hegotá raised a different readiness question: developers postponed an issuance decision while discussing how to make one.

Sepolia’s schedule now has a merged activation record

Nixo merged Parithosh’s testnet-timing proposal on September 15. Stokes’s follow-up sets Sepolia to epoch 353024, October 6 at 21:53:36 WITA. The second merge occurred on September 18 in Bali time. The merged record leaves Hoodi and mainnet activation times undecided. Those dates should not inherit Sepolia’s certainty.

The devnet report adds evidence beyond scheduling. pk910 said all clients followed devnet-11 and that Lido and Optimism reported successful stack deployments. On devnet-8, however, he reported roughly seven-second Besu block imports under TSTORE spam. Besu’s Justin Florentine said two fixes were in progress. The next useful signal is successful retesting under those workloads, alongside operator-ready releases. A smooth upgrade rehearsal alone cannot settle performance readiness.

Clients separate QUIC defaults from MPLEX retirement

ACDC #187 agreed that Glamsterdam releases should enable QUIC by default. The sequencing matters: Justin Traglia’s September 17 comment records that the MPLEX-deprecation specification will merge after Glamsterdam activates on mainnet. This is a release-coordination decision, not evidence that every distributed client binary already implements the agreed defaults. Watch release notes and actual QUIC connectivity before treating the transport transition as complete.

Test corrections remove misleading client failures

Two test repairs made FOCIL results easier to interpret. ilitteri’s simulator correction requires a null inclusionListSatisfied for non-VALID payloads. His PR reports 5,489 affected fixtures across four clients and a selected rerun improving from zero to 49 passes. The bounded rerun demonstrates the repaired assertion; it does not establish a clean full devnet.

A companion fixture-generation repair addresses 127 inconsistent expectations, covering stale blob-gas headers and a block-gas budget too small for a valid empty block. The author requires regenerated fixtures and a release to remove those remaining failures. That makes the next fixture release and full Hive run more informative than the merge count.

Consensus testing had a related correction. On September 16, mkalinin merged eserilev’s fix to FCR attestation-step ordering. The exported vectors placed attestations before the slot tick, causing Lighthouse to reject future-slot input, although the generator applied them afterward. Correct replay order makes comparisons between clients more trustworthy.

The issuance debate needs a decision process

For EIP-8363: Tapered Issuance Burn, Stokes deferred a decision and undertook to propose a process by the next call. Enrico Del Fante asked for dedicated discussions involving people beyond the technical call. The development to watch is a concrete process for stakeholder participation and evaluation. Continued discussion should not be read as acceptance of the proposal’s economics or inclusion in Hegotá.

Sources and further reading