OGIT 1:1 import + ogar-from-schema producer with reverse-emit and SGO

[AdaWorldAPI]

What lands

Three things, one commit:

1. OGIT mirrored 1:1 at a pinned SHA. Full upstream vocabulary
   under vocab/imports/ogit/: NTO (72 domains, 1431 TTLs), SGO upper
   ontology (508 TTLs including 176 canonical verbs), root ogit.ttl,
   SDF JSON config samples. 9.3 MB, bijective with origin.
   Sample-200 random byte-equality check: zero mismatches.

2. ogar-from-schema producer family. Three modules — ttl
   (line-oriented OGIT-dialect parser), ttl_emit (reverse emit,
   semantic bijection), sgo (verb registry parser + emit). Lifts the
   structural arm of a domain from declarative schemas; pairs with
   the existing source-AST producers (ogar-from-rails,
   ogar-from-elixir) which carry the behavioural arm.

3. MARS as the third closed-formal calibration domain (after chess
   and OSM-quadkey). The MARS-Schema XSD oracle (MARSSchema2015.xsd,
   frozen since 2015, version 5.3.8) is vendored at
   vocab/imports/ogit/NTO/MARS/_oracle/ together with the upstream
   extract_classes.py (Python 2 as-is + mechanical Py3 conversion).
   The TTL fixed-enum values must agree byte-for-enum with the
   XSD-extracted classification set — mechanically tested.

Three levels of bijection, all mechanically enforced

Level	What it proves	How
1. Byte equality vs upstream	The mirror is bit-identical to AdaWorldAPI/OGIT at the pinned SHA	diff -qr (manual, documented in calibration/mars/README.md)
2. XSD-oracle agreement	TTL fixed-enum values appear in the XSD-extracted set (chess-grade calibration at the schema layer)	ttl::tests::application_class_values_appear_in_xsd_oracle
3. Semantic round-trip	parse(emit(parse(src))) == parse(src) for every predicate the OGIT dialect uses	ttl_emit::tests::all_mars_ttl_files_roundtrip + sgo::tests::all_sgo_verbs_roundtrip

Whitespace, comment positions, and @prefix declaration order are
intentionally not preserved — they are not load-bearing for the
structural arm, and pursuing byte-bijection would force the producer
to carry raw text alongside parsed structure, defeating the "schema
as IR" pattern.

The finding logged in EPIPHANIES

Three new dated entries (FINDING):

• Schema-vs-source duality. Schemas lift the structural arm
  bijectively (declarative-by-construction). Source ASTs lift the
  behavioural arm best-effort (dynamic dispatch defeats static
  extraction). The two cover disjoint surfaces that meet only at
  the structural arm and become each other's oracle there. Emitting
  a schema from a source-lifted Class and diffing against the
  committed schema is "ontology change management" — what Palantir
  Foundry charges money for, free here from extract_classes.py
  plus 50 LOC of reverse-emit.
• Reverse-engineering bijection. Producer is symmetric:
  anything lifted from OGIT TTL can be emitted back. No two-way
  translation table is needed for bardioc's migration — the
  producer IS the translator.
• SGO is the AST predicate vocabulary. 176 verbs (dependsOn,
  contains, runsOn, generates, relates, causes, …) sitting
  in the upper ontology are exactly the predicates every NTO entity's
  ogit:allowed (…) block references. Lifted into typed VerbDecl
  registry; string-compared references become typed lookups for
  Association / ActionDef.

Docs (reverse-pyramid, pattern density first)

• docs/HIRO-IN-CLASSES.md — bardioc-efficiency story: what changes
  with the lift (dispatch, validation, dependency graph, similarity)
  and what stays (XSD validation, REST surface, audit). Concrete
  byte/ns numbers per call.
• docs/MARS-TRANSCODING.md — the calibration spec. Four-entity
  A→R→S→M backbone, oracle pipeline, six IR-shape tests audit.
• docs/FOUNDRY-ODOO-MARS-LENS.md — three angles on one ontology.
  What MARS teaches Odoo (frozen-schema drift detection); what Odoo
  teaches MARS (_inherit mixin composition); what OGAR adds (IR as
  contract, not vendor platform).
• docs/OGIT-DOMAIN-LIFT-CATALOGUE.md — coverage register for the
  72 NTO domains with Imported / Lift-tested / Cross-walked /
  Production status. Avoids future re-fetch and re-derivation cycles.
• docs/calibration/mars/README.md — recipe to re-prove the
  bijection from scratch (three commands).
• docs/DOMAIN-INSTANCES.md — MARS row + capability matrix columns
  for frozen-schema bijection oracle, reverse-emit, AST predicate
  registry.

Provenance

• vocab/imports/ogit/PROVENANCE.md — full OGIT, SHA-pinned at
  d0f489fff94640fef1e6abe7eacba90a1a144579 (2026-05-30), MIT
  licensed.
• vocab/imports/ogit/NTO/MARS/PROVENANCE.md — adds the XSD oracle
  from arago/MARS-Schema with Py2→Py3 conversion documentation.

Test footprint

• 15/15 tests in ogar-from-schema (round-trips + XSD-oracle
  agreement on 29 MARS TTLs + 176 SGO verbs).
• 296/0 workspace passed/failed.
• Clippy-clean (--no-deps) and fmt-clean on the new crate.

Re-vendor recipe

Documented in vocab/imports/ogit/PROVENANCE.md. Three cp -r + one
SHA bump. Future SHA updates are mechanical; the bijection tests
catch every drift.

---

Generated by Claude Code

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.