Unit Tests & Oracle Dashboard
A correctness inventory with unlike units kept separate: reflected unit tests and unit-test classes, committed NumPy 2.4.2 Oracle test cases, specialized flags/layout/format cases, execution gates, and live interoperability suites.
One reflected method declaration; not an executed result
Grouped ownership surface across the reflected assemblies
Corpus plus specialized flags, layout, and NPY/NPZ cases
Classes that run committed Oracle cases and harness checks
Legend & How To Read
Unit test inventory
Unit tests remain summarized by execution policy and capability group; they are intentionally kept out of the Oracle explorer.
Oracle explorer
Browse every Oracle operation by test cases, files, recorded layout labels, dtypes, parameters, result kinds, and explicit errors.
Status bands
Unit Test Execution Policy
Unit-test methods by runner policy; these are inventory counts, not pass/fail results
Unit Test Groups
Source-folder suites deduplicated into broad capability groups; rows show unit tests and classes
Dtype Coverage
Non-exclusive Oracle test case↔dtype links; mixed-dtype cases can contribute to several dtypes
Specialized Oracle Tests
Oracle test cases outside the common operation/index corpus schema
Oracle Evidence Explorer
Oracle operations and their committed test cases; unit tests stay in the aggregate inventory above
Coverage Review Queue
Factual evidence counts first; applicability-aware expansion techniques second
Full Reports
Generated data and the human-maintained oracle ledgers
- Full unit test inventory CSV — one row per reflected unit-test method.
- Generated inventory summary — headline counts and strength queue.
- Oracle architecture and divergence ledger.
- Coverage map and expansion techniques.
- Journey3 touched-function receipt.