Remove the dead execute_parallel and execute_sequential methods

[Nitjsefnie]

Summary

Removes Executor.execute_parallel and Executor.execute_sequential, which nothing in the package calls, along with the two unit tests that were their only callers.

Related Issues and Pull Requests

Fixes #52

Changes

• openagent_eval/core/executor.py — deletes both methods.
• tests/unit/test_core/test_engine.py — deletes test_execute_parallel and test_execute_sequential.
• CHANGELOG.md — entry under [Unreleased] → ### Removed, following fix: use metadata parameter in ChunkingQualityAnalyzer (#65) #204's shape.

Chose deletion over the TODO marker your issue also offered. The precedent is this repo's own: #75 and #77 were both dead-code findings resolved by deletion and merged. There's no TODO-marker precedent to follow, the methods appear in no docs, examples, changelog or release notes, and the package is pre-1.0 with no deprecation convention. Happy to switch to markers if you'd rather keep them as an intended API.

Testing

One correction to the issue worth stating plainly: "never called anywhere" isn't quite right — they had two dedicated unit tests. No production caller exists, which is what makes deletion correct, but tests/unit/test_core/test_engine.py called them at four points, so those tests had to go with the methods or the suite would AttributeError. That's why the unit count drops by exactly two.

Verified as an enumeration rather than a spot check. Post-change, grep -rn "execute_parallel\|execute_sequential" over the whole tree — source, tests, docs, examples, scripts, config — returns exactly one hit: the CHANGELOG entry itself. A hasattr check on the class confirms both are gone.

gate	result
pytest tests/unit	1013 passed, 4 skipped (baseline 1015 — the two deleted tests)
full suite w/ coverage	1067 passed, 4 skipped (baseline 1069)
coverage	81.18%, threshold 75. executor.py goes 55 stmts/85% → 33 stmts/100%
mkdocs build --strict	pass
python -m build + twine check	pass

The soft gates add no new findings — checked at finding level against a base worktree rather than by comparing totals: ruff 224 vs 228 at base, format clean, mypy unchanged, integration 54 passed.

Footer

Generated by Claude Opus 5 (review), Kimi K3 (brief, implementation, verification)

[himanshu231204]

/oc review pr

[opencode-agent]

PR Review: Remove dead execute_parallel and execute_sequential

Verdict: Approve. The change is correct, minimal, and well-scoped.

Verification performed

• Read both commits (7d04db1, ab5ba9a) in full.
• grep across the whole tree — the only surviving references are in the CHANGELOG entry (expected).
• Confirmed the Coroutine import removal is safe: grep Coroutine in executor.py now returns nothing, so it was only used by the deleted methods' type hints.
• Confirmed the two deleted tests are exactly the parallel/sequential tests and nothing else in the suite references the removed methods.

Assessment

Correctness — The two removed methods were redundant with Executor.gather() (the pipeline's real path). Deleting them plus the now-orphaned Coroutine import is consistent. The PR body correctly corrects the issue's claim: they were tested (in tests/unit/test_core/test_engine.py), which is precisely why the tests had to be removed too, avoiding an eventual AttributeError. The test-count math (1015→1013) matches.

Consistency with conventions:

• Matches the repo's own precedent (metrics(claim_extractor): _SIMPLE_PATTERNS regex compiled but never used (dead code) #75, providers(anthropic): orphan LLMResponse constructed but never used in generate() #77) of using deletion over TODO markers for dead code — agreed given the issue is unreleased pre-1.0.
• CHANGELOG entry follows Keep-a-Changelog "### Removed" shape and links execute_parallel and execute_sequential are dead code in executor.py #52. Putting something under [Unreleased] → ### Removed before the --- separator is correct.

Minor nits (non-blocking):

• CHANGELOG entry is quite verbose (backticks, prose, parenthetical). The repo's other entries use a terse **Title** — summary (#ref) format, so consider trimming the parenthetical, e.g. remove Executor.execute_parallel() and execute_sequential() (no production callers; see gather) (#52).
• No test environment was available in this sandbox (uv/pytest not installed) to run the suite myself; the PR author's self-reported gates (1013 passed, executurutils 100%) are consistent with the small diff, but I couldn't independently rerun.

Both edges (import removal, test removal, changelog) are correct. I'd ship it.

opencode session  |  github run

[github-actions]

🎉 Congratulations @Nitjsefnie!

Your pull request has been successfully merged into main. 🚀

Thank you for contributing to OpenAgentHQ and helping improve the project.

We truly appreciate your contribution and hope to see you back with more amazing PRs!

Happy Open Sourcing! ❤️