W2.B.4 — fanout reducer race test + deterministic merge#55
Draft
TimothyVang wants to merge 2 commits into
Draft
Conversation
added 2 commits
May 2, 2026 08:44
Adds tests/graph/test_fanout_race.py exercising the LangGraph fanout through the canonical Send-based parallel-edge idiom: - 4 branches (vol/hay/pls/mft_exec), each sleeping 0-500ms, must merge in DETERMINISTIC order (sorted by branch_name then sequence_id) so the HMAC ledger chain hashes to identical bytes across runs; - two runs with different random seeds must produce identical executor_results orderings; - no branch is silently dropped (ARCHITECTURE.md §1 empty-set rule); - reducer is idempotent across .ainvoke() calls (no cross-run accumulator state). Plus pure-reducer unit tests: - None left/right -> [] - (branch_name, sequence_id) sort key - keyless items sort to tail in stable insertion order. This commit also REGRESSES verdict/graph/reducers.py to a naive concat (no sort) so the test suite goes RED first per CLAUDE.md §3.7 TDD discipline. The W2.B.1 reducer was speculative-design; this commit pins the contract via failing test, GREEN restores the deterministic sort.
Restore the deterministic merge in verdict/graph/reducers.py: - Sort key is (branch_name, sequence_id); items missing those keys use chr(0x10FFFF) so they sort to the tail in stable insertion order; - Python's stable sort guarantees identical merged bytes regardless of branch completion timing; - None-left/None-right are tolerated (first write, empty update); - HMAC ledger chain reproducibility is preserved across runs even when wall-clock branch order differs (the failing case in the W2.B.4 RED commit). Tests: 31 passed (W2.B.1 + W2.B.2 + W2.B.3 + W2.B.4). Ruff: clean.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
tests/graph/test_fanout_race.pyexercising the LangGraph fanout through the canonicalSend-based parallel-edge idiom against a real compiled graph (no mocks).vol_exec/hay_exec/pls_exec/mft_exec) each sleep 0-500ms and must merge in deterministic order; two runs with different RNG seeds must produce identicalexecutor_resultsorderings (HMAC ledger chain reproducibility, ARCHITECTURE.md §2).verdict/graph/reducers.pyafter the W2.B.4 RED commit regressed it to a naive concat — the test catches the regression by failing on both wall-clock-race and pure-reducer paths.Nonehandling,(branch_name, sequence_id)sort key, keyless items sort to tail in stable insertion order.test(graph): RED — fanout reducer race test [W2.B.4](3 fails) → GREENfeat(graph): reducer pattern for parallel-executor merge [W2.B.4](31 pass).Test plan
pytest tests/graph/test_fanout_race.py -v— 6 passed.pytest tests/graph/ -v— 31 passed.ruff check— clean.Base
This PR targets
feat/W2.B.3-comprehension-mismatch-ledger.