feat(parity): #259 foundation — Layer 2 normalized lowered representation - #299
Merged
Conversation
…tion
The observability surface the Rust own-bridge port is gated on
(spec/Bridge.md §6, layer 2): a Python-authored canonical JSON projection of
what to_module() lowered — Module AST + normalized handle map — taken BEFORE
any analysis runs, so a wrong lowering is visible on its own instead of
hiding behind an unrelated silence at the diagnostics layer.
* ownlang/lowered.py — the emitter (read-only observer; imported by nothing
on the production verdict path). The docstring FREEZES the normalization
decisions: fixed field order; semantic (unsorted) array order per BR-D4;
closed statement shapes under a `stmt` discriminator with no optional
fields (null only for a bare return); verbatim conds and callee strings;
prelude/externs always, lifetime declarations exactly when minted; a
normalized handle-map allowlist in fixed order (never a copy of the facts
document); OwnIRError projecting as {"error": ...}; LOWERED_VERSION keys
the surface. Unprojectable AST shapes fail loud — the projection must not
silently shadow an unknown lowering.
* tests/fixtures/lowered/ — 26 minimal facts/golden pairs covering the
mandated lowering families: routing R1-R6 (incl. released/conservative
capture skips and the returned_fresh instance-beats-DI rule), global
sub_/cap_ counters, parg_/loc_ ordering, capture lifetime declarations,
DI-routed capture (+ subscriber's own region), the OD-2 tolerant-door
unknown-kind fallback pinned as documented behavior, kill-on-rebind,
unmapped use/release/overspan/bare return, cross-branch hoisting positive
(release and use-only) and the negative gates (nested depth, while-body,
early-return; pool-kind preservation), optimistic in-branch untrack,
top-level kill-site $consume, direct consume call, sig-precise vs merged
overload resolution, unknown-callee drop, first-party + Tier B fresh
mints, alias_join (incl. untracked-src no-claim + rebind), source-line
preservation, and the fail-loud unknown-op rejection text.
* tests/test_lowered_fixtures.py — verify/--write harness in the
cfg_parity mould: stale, missing, and orphaned goldens are each a red
build; a per-case double-render pins determinism; --write regenerates
and prunes orphans. Auto-discovered by tests/run_tests.py.
* spec/Bridge.md §6 + BridgeBehaviorMatrix.md footer reconciled from
"no Python emitter yet" to the built surface.
Guardrails held: no rust/crates/own-bridge, no validation/MOS port, no
ownir.py or schema or diagnostics change (git shows additions + the two
spec files only — production findings are byte-identical by construction),
no CLI change (the emitter is API + test harness). Full suite, ruff, mypy
(strict, now 30 files), and the Rust workspace stay green; regeneration
leaves a clean tree.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MdYFKUaygHz1T9H1qJ7BqK
|
Warning Review limit reached
Next review available in: 54 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (57)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…hapes, discriminating fixtures Five review findings on the Layer 2 surface, each closing a known loophole before the contract freezes: 1. tolerant_unknown_kind is now a PYTHON-ONLY behavior snapshot: the new manifest marks it rust_replay: false with decision: "#294", and the emitter/harness/spec wording now says Rust replays every rust_replay: true case — a Python-only ambiguity case pins an open decision and takes no side on how #294 resolves. The fixture no longer decides OD-1/OD-2 for the port by accident. 2. tests/fixtures/lowered/manifest.json — the frozen case ledger (name, BR/OD rules, rust_replay, decision). The harness requires manifest == facts == goldens EXACTLY, so deleting a facts+golden PAIR (invisible to the per-file checks) is now a red build, and --write refuses to regenerate a shrunken contract (missing facts for a manifest case, or an unlisted facts file, are errors). "Cannot silently rot in any direction" is now technically true. 3. handles are an ordered ARRAY in mint order with `handle` as the first field of each entry — JSON object key order is not a sound carrier of semantic order for a byte-exact cross-language contract (BR-L2/BR-D4 now expressed explicitly, not via map insertion behavior). 4. The closed projection fails loud on every intentionally-unprojected shape instead of silently dropping it: non-empty Acquire.args, module policies, an extern return type, prelude resources carrying emission templates. The docstring also states the Layer 2 boundary explicitly: lowered Module + identity/routing metadata; lambda/ignore_reason are Layer 3 verdict-presentation metadata and excluded deliberately. 5. The overload fixture now DISCRIMINATES channel from kill-site through observable semantics (no invented origin field): after the sig-precise channel $consume the post-call `use c1` survives and projects; after the merged-may kill-site the post-call release vanishes via unmap. A lowerer that wrongly kill-sites the precise call now changes the golden. Negative gates exercised by hand: pair-deletion red, unlisted-facts red, --write refusal, plus the prior stale/missing/orphan trio. Full suite, ruff, mypy, cargo green; double regeneration leaves a clean tree. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MdYFKUaygHz1T9H1qJ7BqK
Owner
Author
|
Independent review verdict (Layer 2 foundation, #259): APPROVED, no blocking findings. Verified on
Non-blocking (recorded, no commit now): the manifest harness does not yet validate that Post-merge status: Layer 2 foundation ACCEPTED / FROZEN at Generated by Claude Code |
PhysShell
marked this pull request as ready for review
July 18, 2026 18:17
11 tasks
PhysShell
pushed a commit
that referenced
this pull request
Jul 19, 2026
The normative documents described the pre-slice-3 world. Recorded now: * spec/Bridge.md §6 Layer 2: #299 (Python emitter + fixture foundation), #300 (typed own-lowered surface/emitter, presence-aware metadata, per-document version gate), #301 (own-bridge constructs Layer 2 from facts and reproduces all 26 shared goldens byte-exact; golden is expected output only); tolerant_unknown_kind stays Python-only under #294 (the Rust bridge fails loud on an unknown kind instead of adopting the tolerant fallback); Layer 1, Layer 3, analysis wiring and #259 as a whole remain open. * spec/BridgeBehaviorMatrix.md: the Layer 2 summary paragraph updated to the same completed state. * tests/test_lowered_fixtures.py docstring: future tense ("will replay", "until the Rust emitter exists") moved to the present. * BR-L7 wording nit folded in (recorded on #297): condition (4) guards the post-merge reference/discharge, not only a release — the use-only hoist fixture is in the gate too. Docs-only; no Rust code, facts, manifest or golden changes. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MdYFKUaygHz1T9H1qJ7BqK
13 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Что и зачем
Foundation slice #259 (строго ограниченный scope): наблюдаемая промежуточная поверхность Layer 2 из
spec/Bridge.md§6 — Python-authored канонический emitter нормализованного lowered representation (ownlang/lowered.py, read-only observer поверхto_module(), ни одним production-путём не импортируется), 26 committed facts/golden пар (tests/fixtures/lowered/), детерминированный regeneration path и CI-gate (tests/test_lowered_fixtures.py, авто-дискавери вrun_tests.py): stale, missing и orphaned golden — каждый красный build; per-case double-render фиксирует детерминизм. Решения нормализации заморожены в docstring emitter'а (fixed field order; semantic array order по BR-D4; закрытые statement-shapes подstmt-дискриминатором без optional-полей,nullтолько для bare return; verbatim conds/callee; prelude/externs всегда, lifetimes ровно когда minted; handle-map allowlist фиксированного порядка вместо копии facts-документа;OwnIRError→{"error": ...};LOWERED_VERSION). Непроецируемые AST-shapes fail loud — поверхность не сериализует «тень». Это НЕ реализация own-bridge: Rust-код не добавлялся.Покрытие матрицы (BridgeBehaviorMatrix)
Строки секции (b), получившие прямой Layer 2 golden:
routing_r1_unresolved,routing_r2_subscribe_self,routing_r3_capture_static(unreleased/released/conservative),routing_r4_returned_fresh(+ unknown-provenance twin),routing_r5_di_capture(+ регион самого subscriber'а из services),routing_r6_token_kinds;handles_global_counters; parg_/loc_ ordering →fn_params_ordering; capture lifetime declarations →routing_r3/r5(и их отсутствие в token-only кейсах);flow_kill_on_rebind,alias_join_cases; unmapped use/release/overspan/bare-return →flow_unmapped_refs;hoist_*(5 кейсов);alias_join_cases; overspan/OWN025 L2-часть (Overspan stmt) →lines_preserved; unknown-callee drop →mos_call_unknown_drop; source lines →lines_preserved; vocabulary fail-loud →vocab_unknown_op(текст отказа — часть поверхности); OD-2 tolerant-door fallback запинен как documented behavior →tolerant_unknown_kind;subscribe) присутствуют в goldens — проверено обходом.MOS-sensitive строки секции (c) с L2 golden: direct consume call (
mos_call_direct_consume), in-branch optimistic untrack (mos_untrack_inbranch), top-level kill-site$consume(mos_killsite_toplevel), sig-precise vs merged overload (mos_call_channel_overload_sig; отмечено: у sig-less merged-may вызова$consume— это kill-site discharge, не channel — оба механизма честно видимы), first-party + Tier B fresh mints (mos_fresh_mint).Отложено с обоснованием:
.own» — это textual twin (to_own), не Module-проекция; twin-rule BR-L1 остаётся запинен существующимиto_own+parse проверкамиtest_ownir.py;_FLOW_OPS-declared-but-unhandled internal inconsistency — недостижимо из данных по построению (это dev-инвариант, не входная форма);summariesdump, INF-R1), не L2.Acceptance packet
project_lowered/render_lowered) + 26 goldens + regeneration (--write, с prune orphans) + stale/missing/orphan failure gates (проверены фактически, все три срабатывают). Rust-only: none. Changed: parity infrastructure only (+ реконсиляция двух строк вspec/Bridge.md§6 /BridgeBehaviorMatrix.mdfooter с «not built yet» на built).git diffvs main: только добавления + эти две spec-правки — production findings побитово идентичны by construction (ownir.py/schema/diagnostics/CLI не тронуты; OD-1–6 не «исправлялись» — OD-2 запинен как есть).python tests/run_tests.py,ruff,mypy(strict, 30 файлов),cargo test(весь workspace) — зелёные; повторная регенерация → чистый git tree.Тип изменения
Как проверено
python tests/run_tests.py(новый тест авто-дискаверится: «lowered (Layer 2) fixtures OK: 26 cases»)ruff check .иmypycargo test— Rust workspace зелёный, не менялсяСвязанные issue
Refs #259 (foundation slice — НЕ закрывает issue), #258, #250, #294 (OD-2 запинен), #296.
Чеклист
Draft — не мержить, не self-accept. Гейт: независимый review поверхности — подтверждение, что она действительно видит lowering (routing-решения, handle identity, hoist/untrack/kill-site, MOS-применение), а не сериализует его тень. Полный Rust own-bridge этим slice НЕ разрешается.
🤖 Generated with Claude Code
https://claude.ai/code/session_01MdYFKUaygHz1T9H1qJ7BqK
Generated by Claude Code