PRODUCTION_READINESS_ROADMAP - TheDaniel166/moira GitHub Wiki

Production Readiness Roadmap

Status: active release-hardening roadmap

Purpose

This document defines the remaining work required to move Moira from production-capable engine state to a disciplined release candidate.

This is not a feature roadmap. It is a release-integrity roadmap.

Moira has already undergone major hardening in:

temporal semantics
house-policy propagation
canonical vessel immutability
transit and occultation semantic honesty
progression and sidereal solver correctness
Hermetic and classical decan engine integrity
API-reference and standards alignment

What remains is the final release layer:

compatibility accounting
documentation closure
broader acceptance validation
specialist-module review completion
release-candidate gating

1. Release Contract

Goal: state clearly what changed and what callers must now expect.

Status: complete as of 2026-04-10

Completed outputs:

release notes: wiki/03_release/RELEASE_NOTES_2.0.0.md
compatibility notes: wiki/03_release/COMPATIBILITY_NOTES_2.0.0.md
versioning decision: major release — version bumped from 1.2.1 to 2.0.0 in pyproject.toml

Behavior changes recorded:

naïve datetime inputs now raise instead of being treated as UTC
several canonical result vessels are now frozen or tuple-backed
invalid policy or invariant violations now raise concrete runtime errors
classify_house_system() now raises on unknown codes
Hermetic decan_at() no longer accepts reader
Hermetic decan night-hour computation now rejects invalid night geometry

Exit criteria met:

release note drafted
compatibility notes written with per-change migration guidance
version bump recorded as major (strict semantic versioning)

2. Documentation Closure

Goal: ensure that repository doctrine and user-facing docs tell the truth about the executable engine.

Status: complete as of 2026-04-10

Sweep completed:

wiki/02_standards/API_REFERENCE.md — version corrected to 2.0.0; all datetime, house-policy, decan, and vessel docs verified against live behavior
wiki/02_services/SERVICE_LAYER_GUIDE.md — three stale moira.Chart-is-mutable claims corrected; Chart schema illustration updated to frozen=True, slots=True
moira.wiki/ — regenerated from sources by scripts/sync_git_wiki.py (40 files written, 79 unchanged)
wiki/02_standards/HOUSES_BACKEND_STANDARD.md — already current; HouseCusps fields, policy enums, and fallback doctrine are all correctly documented
wiki/03_standards/VALIDATION_CASE_STELLAR_HELIACAL_RISING.md — already correctly documents the removal of the elongation magnitude guard

Known dangerous stale phrases confirmed absent:

treated as UTC — not present in any primary doc
stale mutable vessel descriptions — Chart mutability claims corrected in service layer guide
removed parameters such as decan_at(..., reader=...) — not present in any doc; API reference already shows clean signature
AssertionError — not present in primary docs

Doc-consistency check: passes clean.

Exit criteria met:

all known stale contract claims removed from primary docs
root-public versus module-public paths are clearly distinguished in API reference

3. Acceptance Matrix

Status: complete as of 2026-04-10

CI lane

Added .github/workflows/release-acceptance.yml.

Kernel-free slices running in CI on every push and PR:

Slice	File	Cases	Oracle	Note
Hermetic decans	`tests/unit/test_hermetic_decans.py`	87	internal cross-consistency	all pass
Classical decanates	`tests/unit/test_decanates.py`	71	internal cross-consistency	all pass
Sidereal unit	`tests/unit/test_sidereal.py`	51	internal cross-consistency	all pass
Houses external reference	`tests/integration/test_houses_external_reference.py`	2	offline Swiss Ephemeris fixture	all pass
Sidereal external reference	`tests/integration/test_sidereal_external_reference.py`	77	offline swetest fixture	excl. `1625_03_16_ut12` — see gap below

Local-only slices (require DE441 kernel)

kernels/ is git-ignored; DE441 (3.1 GB) is not bundled. These slices are validated locally before release and cannot run in stock CI.

Slice	File	Local result
Eclipse external reference	`tests/integration/test_eclipse_external_reference.py`	pass
Eclipse NASA reference	`tests/integration/test_eclipse_nasa_reference.py`	pass
Occultations external reference	`tests/integration/test_occultations_external_reference.py`	pass
Progressions external reference	`tests/integration/test_progressions_external_reference.py`	skip (auto)

Known gap log

Sidereal 1625 epoch tolerance: test_sidereal_external_reference.py at 1625_03_16_ut12 (all ayanamsa systems) fails with a ~30 arcsec error against the 3.6 arcsec CI threshold. This is a pre-existing accuracy limit at extreme historical dates and is not a 2.0.0 regression. All modern-epoch cases pass. The CI deselect flag -k "not 1625_03_16_ut12" isolates this gap explicitly. Resolution is deferred; the known limit is documented here for audit continuity.

Exit criteria — met

acceptance matrix codified: ✓
results reproducible in .venv: ✓
tolerances documented: ✓
CI lane live: ✓ (.github/workflows/release-acceptance.yml)

4. Remaining Specialist Audit

Status: complete as of 2026-04-10

Modules audited: electional.py, dasha.py, timelords.py, cycles.py

Audit lens applied to each: public-surface honesty, hidden defaults and fallback doctrine, vessel integrity, temporal semantics, validation sufficiency.

Results summary:

Module	Tests	Result	Notes
`electional.py`	9	clean	—
`dasha.py`	66	clean	one deferred observation — see receipt
`timelords.py`	100	clean	—
`cycles.py`	105	clean	—

Deferred observation (dasha.py, non-blocking): Result vessels DashaActiveLine, DashaConditionProfile, DashaSequenceProfile, and DashaLordPair carry MACHINE_CONTRACT "mutable": false but are decorated @dataclass(slots=True) without frozen=True. DashaPeriod requires post-construction mutation for .sub accumulation; the others are immutable by convention. No runtime safety gap; no change required. Documented in the receipt for audit continuity.

Completion receipt: wiki/03_release/SPECIALIST_AUDIT_COMPLETION_2026-04-10.md

Exit criteria — met:

no known protected-zone contradictions remain open: ✓
major findings either fixed or explicitly deferred with rationale: ✓
all 280 combined test cases green: ✓

5. Operational Hardening

Goal: make future regressions harder.

Status: complete as of 2026-04-10

Completed outputs:

CI gate added in .github/workflows/release-hardening.yml
doc-consistency checks added in scripts/check_doc_consistency.py
protected-zone regression discipline recorded in wiki/03_release/PROTECTED_ZONE_REGRESSION_DISCIPLINE.md
completion receipt recorded in wiki/03_release/OPERATIONAL_HARDENING_COMPLETION_2026-04-10.md

Required work:

add CI lane for the release acceptance matrix
add doc-consistency grep checks for known dangerous phrases
require targeted tests whenever protected-zone logic changes

Recommended doc-consistency checks:

treated as UTC
stale mutability claims
stale removed parameters
outdated house-vessel field descriptions

Exit criteria:

CI gate exists
doc-consistency checks exist
protected-zone regression discipline is documented

6. Release Candidate Gate

Goal: define a real stop/go boundary for production release.

A release candidate should not be cut until all of the following are true:

targeted unit matrix passes
release acceptance matrix passes
primary docs reflect current behavior
compatibility notes are written
versioning decision is recorded
no unresolved protected-zone contradiction is known
root public exports in moira.__init__ have had final review

Exit criteria:

release-candidate checklist signed off

7. Recommended Sequence

Draft release notes and compatibility notes.
Complete the remaining documentation sweep.
Formalize the acceptance matrix in CI.
Audit the remaining specialist modules.
Cut a release candidate.
Run one final regression and acceptance pass.
Release.

8. Current Assessment

Current state:

engine integrity is substantially stronger than before the audit passes
high-risk temporal and semantic defects have been removed from core subsystems
public API truth is much better aligned with runtime behavior
major doctrinal engines now have better boundary discipline

This supports the following practical assessment:

Moira is near release candidate quality at the engine level, but should still complete the final release-contract, documentation, and acceptance steps before being described without qualification as fully production ready.