docs: append task 2 re-review report

2026-09-22 18:24:08 -03:00
parent 2ee271ca80
commit 25af5cf0f0
1 changed files with 57 additions and 0 deletions
@@ -77,6 +77,63 @@ Implementation commit: `d729710` (`feat: add bounded operational view models`).
 - `BlockedProduct` has no persisted installation foreign key. Blocker list/detail identity therefore uses the lowest-ID installation for the tenant as the representative context; agent and installation filters check tenant installation membership without duplicating blocker rows.
 - The existing non-failing pytest environment warnings remain as noted above.

+## Task 2 Re-Review Fix Report
+
+### Findings Fixed
+
+- Unified fiscal completion rounding with deterministic one-decimal `ROUND_HALF_UP` behavior in Python and SQL. The SQL ratio is explicitly numeric before two-argument `ROUND`, preserving SQLite/PostgreSQL compatibility.
+- Extended `blocker_fingerprint` with an optional installation identifier. Legacy calls without an installation retain the original hash; authenticated upserts scope both generated and supplied fingerprints to the exact installation, so identical reports from two installations remain distinct.
+- Normalized blocker attention candidate fields with SQL `UPPER`, matching the normalized breakdown behavior for lowercase `cfop`, `st`, and other required fields.
+
+### Re-Review RED Evidence
+
+Command after adding the three regression tests:
+
+```text
+python -m pytest central/tests/test_operational_view_models.py::test_worst_coverage_sort_cannot_disagree_with_one_decimal_display central/tests/test_operational_view_models.py::test_lowercase_required_blocker_field_reaches_attention_candidates central/tests/test_blocked_products.py::test_authenticated_installations_get_distinct_fingerprints_for_identical_blockers -q
+```
+
+Result before implementation: `3 failed`.
+
+The failures reproduced the SQL/Python 6.3 versus 6.2 ordering mismatch, omitted lowercase blocker attention, and deduplication of identical blocker reports across installations.
+
+### Re-Review GREEN Evidence
+
+Focused Task 2/control/blocker tests:
+
+```text
+python -m pytest central/tests/test_operational_view_models.py central/tests/test_control_plane.py central/tests/test_jobs.py central/tests/test_blocked_products.py central/tests/test_operational_state.py -q
+```
+
+Result: `37 passed`.
+
+Full central suite:
+
+```text
+python -m pytest central/tests -q
+```
+
+Result: `256 passed, 1 warning`.
+
+The warning remains the pre-existing `openpyxl` print-area warning; the environment also emits the existing `pytest-asyncio` fixture-loop-scope deprecation warning.
+
+PostgreSQL SQL compatibility check:
+
+```text
+python -c "from sqlalchemy import select; from sqlalchemy.dialects.postgresql import dialect; from central.app.control_plane import _fiscal_completion_expression; print(select(_fiscal_completion_expression()).compile(dialect=dialect()))"
+```
+
+Result: the expression compiles as numeric SQL using two-argument `ROUND`.
+
+### Re-Review Fix Commit
+
+Implementation commit: `2ee271c` (`fix: close remaining task 2 review findings`).
+
+### Re-Review Concerns
+
+- Legacy no-installation payloads intentionally retain their original fingerprint and tenant-level behavior; authenticated installation-scoped payloads use the new namespace.
+- The existing non-failing pytest environment warnings remain as noted above.
+
 ## Task 2 Review Fix Report

 ### Findings Fixed