Greptile Summary

Adds a Windows-specific fallback in loadSqliteVecExtension for both the src/ and packages/ copies of sqlite-vec.ts: when sqliteVec.load(db) fails and the resolved extension path ends with .dll, the code retries by stripping the suffix so SQLite can append it itself — matching behaviour already expected on Linux/macOS. The fix is logically correct and the two files are kept in sync.

Confidence Score: 5/5

Safe to merge; the fallback logic is correct and non-breaking on all platforms.

Both changed files apply an identical, well-scoped retry that only activates on Windows when the bundled load path ends with .dll. User-supplied paths are untouched. The only finding is a P2 diagnostic improvement (chaining the original error on double-failure), which does not affect correctness or runtime behaviour.

No files require special attention.

This is a comment left during a code review.
Path: packages/memory-host-sdk/src/host/sqlite-vec.ts
Line: 38

Comment:
**Original error silently lost on retry failure**

If `params.db.loadExtension(extensionPath.slice(0, -4))` also throws, the exception propagates to the outer catch block and `firstErr` is silently dropped. Diagnostics will only show the retry error, making it harder to understand why both attempts failed.

Consider chaining the original cause:

```ts
// Replace the retry call to chain errors:
try {
  params.db.loadExtension(extensionPath.slice(0, -4));
} catch (retryErr) {
  const err = new Error(`sqlite-vec retry without .dll suffix also failed`, { cause: retryErr });
  (err as any).firstError = firstErr;
  throw err;
}
```

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: src/memory-host-sdk/host/sqlite-vec.ts
Line: 38

Comment:
**Original error silently lost on retry failure**

Same issue as in `packages/memory-host-sdk/src/host/sqlite-vec.ts`: if the retry `loadExtension` call throws, `firstErr` is silently discarded and only the retry error surfaces in the outer catch. This makes debugging dual-failure scenarios harder.

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "fix: retry sqlite-vec load without .dll ..." | Re-trigger Greptile}

CI note + review feedback addressed

Failing checks (checks-node-extensions, checks-node-extensions-shard-6): these are pre-existing timeout flakes in extensions/bluebubbles/src/catchup.test.ts — completely unrelated to this PR's changes to sqlite-vec.ts. The same failures appear in unrelated PRs on this branch at the same time.

Review feedback addressed in the follow-up commit (c485f5ad):

• Codex P1 — Persist suffixless path after Windows retry: loadSqliteVecExtension now returns loadedPath (the suffix-stripped path) instead of the original extensionPath after a successful Windows retry. manager-sync-ops persists the returned path and passes it back on subsequent loads (after reindex/reset), so subsequent calls go through the resolvedPath branch with the correct path and no fallback needed.

• Greptile P2 — Original error silently lost: when the suffix-stripped retry also fails, both errors are now surfaced via a combined Error with { cause: firstErr }:

  sqlite-vec: both load attempts failed on Windows. Retry error: <retry>, cause: <firstErr>
  

Codex review: needs maintainer review before merge.

Summary
The branch updates the memory-host sqlite-vec loader to retry Windows bundled .dll loads with a suffixless path, adds loader regression tests, and adds an Unreleased changelog fix entry.

Reproducibility: yes. current main's source path calls sqlite-vec's suffixed Windows load path without retry, and the PR body provides Windows 11 before/after terminal output for the failure and fixed behavior. I did not independently run the Windows scenario in this read-only review.

Real behavior proof
Sufficient (terminal): The PR body includes copied before/after Windows terminal output showing openclaw memory index and openclaw doctor after the fix.

Next step before merge
No repair lane is needed; the latest branch addresses the prior mechanical blockers and is ready for normal maintainer validation.

Security
Cleared: The diff changes an existing local sqlite extension load fallback plus tests/changelog only; it adds no dependencies, workflows, secret handling, downloaded artifacts, or broader permissions.

Addressed all review feedback (clawsweeper P2/P3) in the latest push. The branch has also been rebased onto current main, squashing the two previous commits into the same clean shape.

Clawsweeper P2 — Honor explicit extensionPath before importing sqlite-vec

loadSqliteVecExtension now checks resolvedPath (explicit user-supplied path) and calls db.loadExtension(resolvedPath) before importing the bundled sqlite-vec module — matching the ordering in current main and consistent with the regression test added by BunsDev that mocks the bundled module to throw while still expecting an explicit path to succeed.

Clawsweeper P3 — Changelog entry

Added a single-line entry under Unreleased / Fixes referencing #68892.

Tests added (extends the existing sqlite-vec.test.ts):

1. Windows retry succeeds → ok: true, returned extensionPath is the suffixless path (no .dll)
2. Windows dual failure → ok: false, error message contains "both load attempts failed on Windows" (both errors surfaced)
3. Non-Windows platform does not retry → ok: false, db.loadExtension never called
4. Existing explicit-path test is unchanged and continues to pass

Note on src/memory-host-sdk/host/sqlite-vec.ts

The previous commits touched the src/ copy of this file, but src/memory-host-sdk/host/sqlite-vec.ts no longer exists in main (the package was refactored to a standalone packages/ workspace). The rebase dropped those changes; the fix is now applied only to packages/memory-host-sdk/src/host/sqlite-vec.ts where main expects it.

Re-review progress:

• State: Complete
• Detail: The targeted re-review finished, the durable review comment was updated, and the synced verdict was routed.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/25386881130
• Updated: 2026-05-05T15:55:59.229Z