[dfuse-bot]

Merges upstream Base v1.5.1 into release/1.x. Tag v1.5.1-fh3.1 once merged.

v1.5.1 schedules the Denim hardfork on Base mainnet at 2026-11-05 18:00:00 UTC (1_793_901_600), so mainnet nodes must run this release before then. The merge had no conflicts. streamingfast/reth stays on base-v2.5.2.8-fh3.1-2 (upstream still pins base-v2.5.2.8) and firehose-tracer on 5.7.0. The Firehose block model is unchanged, so the tag keeps fh3.1. No other StreamingFast repository needs a release first.

Besides the merge:

• A test that the built-in mainnet chain spec resolves the Denim schedule and gives the first Denim block (53_556_127) its millisecond FIRE BLOCK time.
• ACTIVATIONS.md: Denim mainnet schedule, and two Denim notes (millisecond block time vs the new RPC timestampMs fields, and flashblocks stopping).
• CHANGELOG entry below.

CHANGELOG

## v1.5.1-fh3.1

### Changed

* Rebased on upstream Base `v1.5.1`. It schedules the Denim hardfork on Base mainnet at
  2026-11-05 18:00:00 UTC, so mainnet operators must run this release before then.
  `streamingfast/reth` stays on `base-v2.5.2.8-fh3.1-2` and `firehose-tracer` on 5.7.0; the
  Firehose block model is unchanged.
* From Denim on, Base publishes no flashblocks, so a node started with `--firehose-flashblocks-url`
  emits no partial blocks after the activation; full blocks come every 200 ms instead.

Test Execution Summary

Branch bump/base-1.5.1 (release/1.x at v1.5.0-fh3.1-2 + git merge v1.5.1, no conflicts).

Build Status

• cargo test --no-run -p base-execution-firehose -p base-firehose-tests -p base-firehose-flashblocks -p base-reth-node -p base-execution-chainspec -p base-common-genesis -p base-common-chains: PASSED
• cargo build --release --package base-reth-node --bin base-reth-node --package base-consensus --bin base-consensus (same command as Dockerfile.sf): PASSED (10 min)
• cargo clippy -p base-reth-node --tests: PASSED, no warnings in bin/node (existing doc-backtick warnings in base-execution-firehose and base-execution-flashblocks are unchanged)
• Smoke run: base-reth-node node --chain base --datadir <empty dir> --disable-discovery for 90 s: PASSED. Prints FIRE INIT 3.1 reth base-1.5.1 and the genesis FIRE BLOCK 0, lists Denim @1793901600 in the hardforks, and starts without a Denim schedule error.

Unit Tests

Package	Status	Notes
cargo test -p base-firehose-tests	PASSED	14 tests (prestate replay, EIP-8130 tracing)
cargo test -p base-firehose-flashblocks	PASSED	44 tests
cargo test -p base-reth-node	PASSED	2 tests, including the new mainnet_block_time_switches_to_milliseconds_at_denim
cargo test -p base-execution-firehose	PASSED	no unit tests in the crate
cargo test -p base-execution-chainspec	PASSED	49 passed, 1 ignored (upstream)
cargo test -p base-common-genesis	PASSED	103 tests
cargo test -p base-common-chains	PASSED	52 passed, 1 ignored (upstream)
cargo test -p reth-firehose -p reth-firehose-tests	PASSED	37 tests, run in a streamingfast/reth checkout at base-v2.5.2.8-fh3.1-2 (the pinned tag, unchanged by this bump); cargo cannot test a git dependency with dev-dependencies from this workspace

Integration Tests

• Status: PASSED
• RUSTFLAGS="-C debug-assertions=off" RUST_MIN_STACK=33554432 cargo test -p base-system-tests --test firehose_denim --test firehose_b20 --test firehose_cobalt_b20, Docker-backed:

Test	Status	Time
firehose_b20::b20_transfer_is_traced	PASSED	15.5 s
firehose_cobalt_b20::cobalt_b20_surfaces_are_traced	PASSED	45.3 s
firehose_denim::denim_activation_block_traces_base_time_install	PASSED	50.6 s

Battlefield Tests

• Status: PASSED
• Pass/Fail: 78 passing, 24 pending, 0 failing
• streamingfast/battlefield-ethereum at ffd9260 (origin/HEAD), op-reth-devnet suite with this branch's release base-reth-node (1.5.1) as the Firehose node: ./scripts/optimism/run_optimism_devnet.sh, then OP_RETH_BINARY=<base>/target/release/base-reth-node RUST_MIN_STACK=33554432 ./scripts/run_firehose_op_reth_devnet.sh, then pnpm test:fh3.0:op-reth-devnet. All goldens matched; no golden changed.
• The 24 pending tests are gated on forks or features this devnet does not have (Amsterdam/BAL, blobs, Arbitrum and BSC specifics, the Transfer-log and millisecond timestamp tests), plus the RPC comparison test, which skips because the devnet finalizes at head − 200 and nothing was final yet. I ran that comparison by hand once the blocks were final: fireeth tools compare-blocks-rpc -p localhost:8089 http://127.0.0.1:28545 1 160 gave 160 identical, 0 different (the tests wrote up to block 156).
• No ERROR or panic in the node log.
• The devnet runs the OP Stack forks up to Isthmus from genesis, not Base's own upgrades (Azul, Beryl, Cobalt, Denim, Everest). Denim's Firehose effects are covered by firehose_denim (see Integration Tests) and by the unit test above for the mainnet block time.

Firehose bug patterns

1. Serial order: the txpool and forwarding changes (batch validation against one state provider, lazy lane invalidation, condition partitioning, opt-in --tx-forwarding-batched-insert) stay on the mempool and builder side and never reach the traced executor. The precompile calldata change only borrows the input instead of copying it. None found.
2. Emit after accept: no change to crates/execution/consensus or the engine validation path in the range. None found.
3. Activations: Denim is now scheduled on Base mainnet at 1_793_901_600 (2026-11-05 18:00 UTC). Its block content effects are traced since v1.5.0-fh3.1. The mainnet schedule resolves (first Denim block 53_556_127 at 1_793_901_601), so the millisecond FIRE BLOCK time applies on mainnet. New finding: Denim ends flashblocks, so the Firehose flashblocks partial-block stream goes quiet after activation. Everest and Zenith are unchanged. See ACTIVATIONS.md.
4. Fields: no change to crates/common/consensus or crates/common/rpc-types in the range. None found.

Open questions

• Firehose flashblocks after Denim. From Denim the sequencer builds with the basic builder and publishes no flashblocks, so a node started with --firehose-flashblocks-url emits no partial blocks after activation (already the case on Base Sepolia since 2026-10-21). Nothing breaks: the extension keeps its connection and only records canonical heads. We can drop the flag after activation in next versions.