VED-002 Sayanadi execution and validation receipt

Date: 8 October 2026. Baseline: engine 77d34158cef7c936fdf6613f9be18ba5e8fad5aa, Moira 6.9.9; the validated implementation is included in the containing authorized source-publication package. Contract: Sayanadi admission standard. Authority: source research and executed plan.

1. Result and implementation ownership

The selected BPHS within-sign Navamsa-ordinal profile is complete locally through standalone calculation, strict name/clock inputs, optional chart composition, nine source-defined bodies, curated Python/facade and typed REST, including reader-bound birth/sunrise composition. VED-002 is LOCAL_COMPLETE for that contract; VED-021/022 follow-through is complete at this package's validation/documentation scope. Publication/release/deployment are separate.

Owner	Changes
moira/avasthas.py	Exact binary64 cell ownership; strict bodies, required Moon, typed names/ghati/policy/context; consistent intermediate trace; explicit optional chart evaluation and separate node results; preserved legacy construction with unknown provenance.
moira/sayanadi_dated.py	Aware birth/location preflight, previous/next sunrise discovery and root brackets, uncertain/unavailable outcomes, reader-bound epoch/Lagna/true-node composition and typed resource/coverage errors.
moira/sayanadi_effects.py	Reconciled attribution prose: legacy conditional summaries are not source-certified quotations or evaluated rules. Existing strings retained.
Root, facade, vedic, _facade_vedic	Fifteen source-owned exports with shared identity and three facade methods; borrowed serving reader used for birth composition.
Server models/router/service/serializer/errors	Discriminated strict clocks; direct/chart/birth requests and typed status/trace/source views; canonical projection and normal preflight/resource/coverage envelopes.
Focused unit/server/integration tests	Corrected primary fixture, independent rational boundaries, name/body/clock/input/omission coverage, analytical uncertainty, real DE441 frames and reader lifecycles, export/HTTP parity and OpenAPI.
Canonical standards/reference/register and generated wiki	Selected contract, migration, dated source limits, executed plan and closure status; generated through owning scripts.

No astronomical/native algorithm, dependency, version, golden fixture, tolerance, test harness or known-issue policy changed. Website/Urania and unrelated state-repository work remain outside the implementation. The plan admitted one executing BPHS profile; the degree-based variant remains its identified optional research frontier, not an unfinished implementation gate.

2. Numerical and source evidence

The Santhanam Sun fixture now uses 7 degrees 12 minutes Taurus and Sa=4, with source Moon-star/Lagna-sign categories represented by visibly synthetic exact longitudes. It asserts total 51, index 3, first remainder 1, second remainder 0, Netrapani/Vicheshta. The old 38-degree/name-1 case is separately synthetic: first remainder 10, same final substate. The unsupported assertion that the printed arithmetic needed correction is removed.

Independent component cases include Hora Ratnam's Navamsa-commentary Mars main sum 57/index 9 only, Rao's Mercury within-sign Navamsa ordinal 7, and source ghati rounding 20g2v -> 21, 30g33v -> 31, 42g30v -> 43. The degree-based Sun example is not used as a Navamsa oracle. All 35 literal sound-table tokens, all five name values and all nine body constants are checked against independently specified expectations.

The exact rational boundary test covers nine subjects times 108 global Navamsa boundaries times lower/base/upper representable inputs times five name values: 14,580 cases, zero component/state/substate disagreements. The unmodified baseline had 1,015 state/substate disagreements in that scope. The reference uses Fraction.from_float of the actual supplied value; Moon-subject cases also account for Moon owning the birth star. Integer trace stages, exact 10-degree ownership, wrap behavior, zero state remainder, ghati-second endpoints and hostile inputs are covered. No epsilon or relaxed tolerance was introduced. This is source-formula/invariant verification, not an external-software or predictive oracle.

3. Composition, transport and real-resource evidence

Analytical tests isolate previous-sunrise day/night and exact-zero ownership, ordinal ambiguity across a sunrise root bracket, typed polar/missing outcomes, offset/zone identity, nonexistent ZoneInfo wall times, independent node evaluation/affliction controls and reader-override restoration. HTTP cases exercise canonical direct/chart/birth parity, all three clock shapes, hostile coercions/identities/partial contexts, unavailable output and distinct resource and coverage errors. OpenAPI retains discriminated clocks and nullable trace/status fields.

Real integration used the actually discovered DE441 SPK, explicit borrowed SPK readers and discovered/configured server startup engines. Four date/frame cases cover India and both US daylight-saving transition dates, Lahiri/Raman ayanamsas, all three named sunrise definitions and optional true-node subjects. Reader-bound UT1/TT/TDB, kernel/ayanamsa receipts, independent rational ghati bounds and direct-state parity pass. The existing VED-015 PAC 22 September 2026 central-station sunrise comparison retains its original 60-second minute-resolution gate and yields ghati ordinal 7 for the selected birth component case. This is not a published full Sayanadi chart.

Tromso's 21 June polar sunrise absence yields typed unavailability, no epoch and no fabricated chart. Explicit readers stay open while borrowed and close under their owning context; startup-engine readers stay open through HTTP and close in the owning test's teardown. No new reader-ownership lifecycle was introduced by composition. The prior active reader context is restored.

4. Actual execution receipts

All execution used C:/dev/moira/.venv/Scripts/python.exe, Python 3.14.3, Moira 6.9.9; native import resolved C:/dev/moira/moira/_moira_native.cp314-win_amd64.pyd. Test flags:

$env:MOIRA_TEST_MODE='1'
$env:MOIRA_NO_DOWNLOAD='1'
$env:MOIRA_STRICT_KNOWN_ISSUES='1'

The known-issue registry is empty. The test harness denied external network; HTTP tests were marked loopback. Commands and XML receipts:

.\.venv\Scripts\python.exe -m pytest tests/unit/test_sayanadi.py tests/unit/test_sayanadi_admission.py tests/unit/test_sayanadi_dated.py tests/unit/test_avasthas.py tests/unit/test_vedic_surface_completeness.py tests/server/test_server_sayanadi.py tests/server/test_vedic_direct_input_hardening.py -m "not external_network" -q --tb=short --junitxml=C:/dev/outputs/ved002-sayanadi-research-2026-10-08/final-focused.xml
.\.venv\Scripts\python.exe -m pytest tests/integration/test_sayanadi_ephemeris.py -m "not external_network" -q --tb=short --junitxml=C:/dev/outputs/ved002-sayanadi-research-2026-10-08/integration.xml

269 focused passes plus nine integration passes: 278 distinct latest passing tests, zero failures/errors/skips. Scoped export/offset checks were repeated after final import curation; they overlap the total and are not counted again. The integration resource receipt reports 12 admitted planetary resource uses, zero skips/failures and DE441 content identity.

Initial implementation tests exposed a misconfigured synthetic altitude fixture and an incorrectly constructed synthetic coverage exception; those fixtures were corrected to the existing substrate contracts. The old export test explicitly excluding unadmitted Sayanadi was updated to verify its newly admitted owning identity. No numerical acceptance budget was changed to pass.

Private receipts/scripts are in C:/dev/outputs/ved002-sayanadi-research-2026-10-08/. The standard's supplied JSON request was extracted and executed through the actual HTTP route: 200, Netrapani/Vicheshta, total 51 and remainders 1/0, with exact facade-response parity (standard-json-smoke.json). The final export/offset rerun passed all 17 overlapping tests with no failures/errors/skips (export-check.xml).

Documentation completion commands passed:

.\.venv\Scripts\python.exe scripts/check_doc_consistency.py
.\.venv\Scripts\python.exe scripts/sync_rest_api_reference.py --check
.\.venv\Scripts\python.exe scripts/sync_git_wiki.py --check
git diff --check
git -C moira.wiki diff --check

The REST reference now records 484 registered paths/operations. An independent project-Python link/register check passed 159 local file/anchor links across eight canonical documents, 24 stable work IDs, seven closed and 17 open/bounded packages, and five optional candidates (completion-documentation-check.json). All 20 changed/new Python files pass Python 3.10 source grammar. Differential Ruff checks against the unchanged HEAD source found zero new diagnostics; 66 pre-existing diagnostics remain: 60 E402 in root imports, five F811 in the facade and one F841 in another avastha family (lint-grammar.json). No unrelated lint cleanup was applied. These are scoped verification receipts, not a full-suite or whole-engine certification.

5. Limits and completion meaning

The selected source recipe and engineering admission are complete locally. The competing degree-lineage quantization/source questions, complete textual certification of the retained effect summaries, mean-node birth composition, terrain/observational sunrise accuracy and predictive-effectiveness studies are outside this bounded contract. Their absence does not conceal unfinished work in the admitted calculation/transport. The output labels legacy prose as not_source_certified with conditions_evaluated=false.

Strict invalid-input rejection, corrected partition-boundary outputs and additive JSON metadata are intentional compatibility changes. Valid existing four-family results and no-Sayanadi omission behavior are preserved. After local validation the user authorized committing/pushing the generated wiki first and the parent engine package second; see register section 24. Package release, deployed endpoints and website adoption remain separate actions.