Skip to content

test: forward HNSW sidecar failure and recovery witness - #7

Draft
RecursiveIntell wants to merge 1 commit into
test/search-replay-cfg-forward-20260928from
test/sidecar-recovery-forward-20260928
Draft

RecursiveIntell wants to merge 1 commit into
test/search-replay-cfg-forward-20260928from
test/sidecar-recovery-forward-20260928

Conversation

@RecursiveIntell

@RecursiveIntell RecursiveIntell commented Sep 28, 2026 •

Copy link
Copy Markdown
Owner

Scope

Dependent on draft #6. Forward-sync tests/storage_lifecycle.rs from the editable Libraries owner at 0b099ec416de60f6adafb182f1d1e83795d05c56; the complete test-file blob matches that owner. The new test deliberately obstructs the local HNSW graph-file path, verifies the SQLite fact remains readable and the pending index operation records a failed attempt, then removes the obstruction and verifies an explicit flush clears pending work and writes the graph file. Add paired-root CI commands for strict Clippy and the four storage_lifecycle tests under hnsw-only. No production-source change.

Evidence

  • A disposable paired root with the exact owner test blob passed the focused new test (1), all four lifecycle tests, strict hnsw Clippy, and formatting.
  • An exact committed-source assembly at this PR head passed the scratch lock-delta check, strict hnsw test-target Clippy, all four lifecycle tests, and formatting. The paired root is not a claim that the raw standalone checkout builds.
  • Hosted paired-root CI run 36492894578 completed successfully at this PR head.

Limits and rollback

This is a deterministic local sidecar-write obstruction with same-process explicit recovery. It does not simulate a process crash/restart or prove ClaimLedger append-to-projection recovery, broad durability, whole-mirror parity, or release readiness. Roll back by closing this dependent draft or reverting the exact two-file commit; do not copy dirty overlays, merge, or activate by implication. Public Evidence Editor and LongMemEval reviewed the narrow candidate at /home/sikmindz/semantic-parity/semantic-memory-sidecar-review-20260928/.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant