MUHURTA_LAGNA_VALIDATION_2026 10 09 - TheDaniel166/moira GitHub Wiki
Muhurta Lagna validation receipt - 9 October 2026
State: selected VED-010 package complete and included in the authorized
wiki-first source-publication package.
Baseline engine 571788a1fe705b096d5a69911f30be583c7c45d7 / generated wiki
ed68937510e6ce162da82420b4f42415c0480161 is the published VED-009 package.
That baseline's hosted Release Hardening and Acceptance Matrix passed. The
results below are local verification of the subsequent VED-010 work.
The source record identifies five scan hashes, inspected pages, source variants and declared numerical choices. The standard defines the admitted public/REST contract and compatibility changes.
1. Implementation and source evidence
moira/muhurta_lagna.py: separate general and marriage Lagna policies; both MC Navamsa readings; fractional-aspect support; MC87/92 placement score; raw restrictions with optional MC88 exceptions; complete nullable prerequisite evidence and optional full Shadbala context.moira/muhurta_lagna_dated.py: serving-reader instantaneous composition with UT1/TT/TDB receipt, true ayanamsa and nodes, canonical house-owner Lagna, optional Porphyry Shadbala and explicit unavailable polar geometry. Caller and reader-derived input provenance are distinguished.- Curated root/facade/Vedic exports: eleven owning names, three facade methods. Three strict REST routes under
/v1/muhurta/lagna/preserve the canonical evidence and resource identity. - Shared Varga owner and its Shadbala/Avastha consumers: rational D9 boundaries. Shared Vedic dignity table: Venus-to-Moon enemy correction, preserving the reverse neutral direction. Both corrections are source-backed and separately tested.
- Public API snapshots, REST inventory, canonical/generated documentation and publication manifest are reconciled. The two kernel-free feature test files are now part of Release Hardening.
tests/fixtures/muhurta_lagna_sources.json independently transcribes all nine
house/weight rows, MC76's four printed fractional-aspect examples, MC85's two
printed last-Navamsa examples plus a clearly labeled derived vargottama case,
and all 42 directional BPHS3.55 relationships. All five private scan hashes
were checked against the canonical source record. This is source validation;
no external astrology program was treated as doctrinal authority.
2. Executed verification
Only project .venv Python 3.14.3 was used. Every pytest command ran with
MOIRA_NO_DOWNLOAD=1, MOIRA_TEST_MODE=1 and
MOIRA_STRICT_KNOWN_ISSUES=1. Private XML and deduplication receipts are in
C:/dev/outputs/ved010-research-2026-10-09/.
| Evidence set | Final outcomes and receipt |
|---|---|
| New source/engine/boundary tests | 315 passed, unit-final.xml. All 108 source house/weight cases; all 109 D9 endpoints with both adjacent floats; node-opposition edges; shared consumers; purpose disagreements; grade endpoints; all fixed-role restrictions; opt-in exception/combustion cases; nullable truth; directional friendship; malformed/stale/nonfinite/negative strength receipts; resource/coverage mapping and context restoration. |
| New REST tests | 41 passed, included in feature-accepted.xml. Kernel-free catalogue/direct parity, strict admission, partial results, typed OpenAPI, full Shadbala transport, error envelopes and facade reader preservation. |
| New DE441 integration tests | 7 passed, included in feature-accepted.xml. Independent planetary and canonical house-angle checks; two ayanamsas, both hemispheres and poles; direct/dated/HTTP parity; configured/discovered/borrowed readers and lifecycle. All 10 requested planetary resource uses ran against content identity DE-0441LE-0441. |
| Existing affected Vedic regressions | 881 passed, deduplicated from acceptance-feature.xml and consumer-regressions.xml, excluding their new-feature cases. Varga/Shodashvarga, Shadbala, Avastha, dignity, yoga, Sayanadi, dosha, Shuddhi and corresponding service/REST consumers. No existing regression failed. |
| Release Hardening original four groups | 1,612 passed, one existing skip, release-gates.xml, 804.874 seconds. All 55 original workflow files ran; the two newly added workflow files are covered by the feature receipts above. acceptance-summary.json verifies membership for all 57 current workflow files across four groups. All 216 DE441 resource uses and the one supplemental small-body resource check ran successfully. |
The final deduplicated outcome is 2,856 passed, one existing skip, zero
unresolved failures/errors across 2,857 test IDs. The new feature accounts
for 363 of the passing tests. Repeated executions are not added to the
total. Earlier development runs contained test-fixture mistakes (enum naming
and a supposed negative case that actually had valid cross-aspects); the
corrected fixtures passed without relaxing the assertions. The latest test
result for every ID is recorded in acceptance-summary.json.
The sole skip is the existing
tests/unit/test_void_of_course.py::test_no_aspect_voc_last_aspect_is_none:
its 30-day scan found no qualifying no-aspect window, so RULE-06 was not
exercised. It is not a VED-010 skip or a missing-resource skip.
Representative exact command set (using the environment above):
.\.venv\Scripts\python.exe -m pytest tests/unit/test_muhurta_lagna.py tests/server/test_server_muhurta_lagna.py tests/integration/test_muhurta_lagna_ephemeris.py -q --junitxml=C:\dev\outputs\ved010-research-2026-10-09\feature-accepted.xml
.\.venv\Scripts\python.exe -m pytest tests/unit/test_muhurta_lagna.py -q --junitxml=C:\dev\outputs\ved010-research-2026-10-09\unit-final.xml
.\.venv\Scripts\python.exe C:\dev\outputs\ved010-research-2026-10-09\run_release_tests.py
.\.venv\Scripts\python.exe -m ruff check moira moira_server scripts --select F401,F821,F822,F823 --no-fix
The release helper records its expanded file list in
release-test-selection.json. The affected-family file selections and all
individual results are recoverable from their XML; repeated new-feature IDs
are superseded by the final passing receipts. All 18 changed Python files
parse under Python3.10 grammar. This is a grammar check, not a claim that the
local suite ran on a Python3.10 interpreter.
All six release-facing artifact checks pass: documentation consistency, release identity, Hellenistic inventories, REST reference, generated Git wiki and website publication manifest. The REST registry has 498 paths/operations (GET40, POST458). Exact public snapshot/docstring governance, full scoped Ruff and whitespace checks pass. Generated Hellenistic family content is unchanged.
3. Admission and remaining boundaries
VED-010 is complete for its named instantaneous composition contract. This does not establish predictive efficacy or universal source agreement. Broader Kartari/general cancellation/KP remedies and continuous strength windows are explicit exclusions, and complete purpose elections remain VED-011. Source availability is not denied. The Raman Sun-minimum discrepancy is recorded; canonical Shadbala thresholds are reused as context, never silently promoted into a Muhurta override.
The legacy Muhurta score/search, Tara/Chandra policy and VED-009 temporal contracts retain their behavior. The shared D9 and Venus/Moon corrections are disclosed compatibility changes. No native substrate, dependency, version, release tag, deployment, website or Urania change is included. Version remains 6.9.9. Existing unrelated state-repository work is preserved; its appended local diary receipt does not publish that repository. The authorized VED-010 source publication includes the generated wiki before the parent gitlink. The results above are local acceptance evidence; hosted checks for the new commit are a separate publication receipt.