Skip to content

feat(corpus): scale OKF conformance 20 -> 33 (W44-B6, C08 L73 close-out) - #368

Merged
KooshaPari merged 1 commit into
mainfrom
feat/sl-w44-b6-corpus
Jul 24, 2026
Merged

feat(corpus): scale OKF conformance 20 -> 33 (W44-B6, C08 L73 close-out)#368
KooshaPari merged 1 commit into
mainfrom
feat/sl-w44-b6-corpus

Conversation

@KooshaPari

Copy link
Copy Markdown
Owner

Summary

Wave-44 close-out lane B6 — production-scale corpus breadth. Scales the OKF
conformance corpus from 20 hand-vetted fixtures to 33 via a deterministic
generator. Closes C08 L73 production-scale corpus breadth residual from
Wave-43 SCORECARD.

What changed

File Lines Purpose
scripts/corpus-generate.py +444 Deterministic OKF fixture generator; OKF_VERSION-keyed; FIXTURE_SPECS + FAILURE_FIXTURES decoupled
docs/reference/conformance/fixtures/*-03[7-9].okf.json +12 aider / opencode / continue / kiro / factory-droid / sql / yaml / large-count / deep-graph / rapid-fire / CJK / embedded-json / multi-modal
docs/ops/corpus-breadth.md +119 Strategy doc; generator API; acceptance; risk register; W45 carry-over plan
tests/replay_breadth.rs +168 5 hermetic tests pinning the W44-B6 deliverable (fixtures on disk, count >= 33, well-formed, generator + docs present)
docs/reference/conformance/README.md +26 Section listing generator-produced fixtures; regenerate instructions

Total: 17 files, +3176/-1.

Coverage gained

  • 5 new agent sources (aider, opencode, continue, kiro, factory-droid-2)
  • 2 new languages (sql, yaml)
  • 3 stress shapes (100 entities, 7-deep relation chain, 12-intent rapid-fire)
  • 3 edge cases (CJK unicode label, embedded JSON in label, multi-modal sha256 hint)

Test wiring

tests/replay_breadth.rs adds 5 hermetic tests:

  1. w44_b6_all_generated_fixtures_are_on_disk — all 13 named slugs present
  2. w44_b6_corpus_total_meets_or_exceeds_target — >= 33 fixtures
  3. w44_b6_each_generated_fixture_is_well_formed_okf_v1 — each parses; okf='1.0'; entities/provenance present
  4. w44_b6_generator_script_present_and_importable — generator on disk; defines OKF_VERSION, FIXTURE_SPECS, FAILURE_FIXTURES
  5. w44_b6_corpus_breadth_doc_present — docs/ops/corpus-breadth.md references C08 L73 and Wave-44

The existing tests/okf_roundtrip.rs::conformance_corpus_fixtures_validate_via_our_parser
automatically picks up the new 13 fixtures (it iterates *.okf.json under
docs/reference/conformance/fixtures/).

Failure-mode fixtures

E01 (malformed-truncated), E02 (missing-provenance), E03 (duplicate-id) live
in a separate scratch dir. They are NOT in the accepted-corpus run; the
harness is expected to reject them. Wiring the negative test is queued for
W45 if needed.

Score disposition

Pre-W44: 396/402 (98% A)
Target post-W44: 402/402 (100% A+); C08 L73 specifically: 29/30 → 30/30
Realistic post-W44: depends on remaining 5 unpaid residuals; see PR #367.

Predecessors

Refs: WAVE44_SCOPE.md (rank 6), docs/ops/WAVE44_PERT.md (lane B6), C08 L73.

- scripts/corpus-generate.py: deterministic Python generator; OKF_VERSION
  keyed; FIXTURE_SPECS (accepted) and FAILURE_FIXTURES (rejected) decoupled.
- 13 new accepted fixtures covering: 5 new sources (aider/opencode/continue/
  kiro/factory-droid-2), 2 new languages (sql, yaml), 3 stress shapes
  (100 entities, 7-deep relation chain, 12-intent rapid-fire stream),
  3 edge cases (CJK label, embedded JSON label, multi-modal sha256 hint).
- docs/ops/corpus-breadth.md: strategy + generator API + acceptance.
- tests/replay_breadth.rs: 5 hermetic tests pinning the W44-B6 deliverable
  (all 13 fixtures on disk, total >= 33, each well-formed, generator +
  docs present).
- docs/reference/conformance/README.md: section listing generator-produced
  fixtures.

Total corpus: 20 hand-vetted + 13 generated = 33 accepted fixtures.
Failure-mode fixtures (E01-E03) live in a separate scratch dir and are
expected to be rejected by the harness.

Refs: WAVE44_SCOPE.md (rank 6), docs/ops/WAVE44_PERT.md (lane B6),
C08 L73.
@coderabbitai

coderabbitai Bot commented Jul 24, 2026

Copy link
Copy Markdown

Warning

Review limit reached

@KooshaPari, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 58 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

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 configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: c9e0c1a0-d160-4de0-97cf-cfab01402c4c

📥 Commits

Reviewing files that changed from the base of the PR and between 513c23c and 0b46ecb.

📒 Files selected for processing (17)
  • docs/ops/corpus-breadth.md
  • docs/reference/conformance/README.md
  • docs/reference/conformance/fixtures/aider-rust-refactor-037.okf.json
  • docs/reference/conformance/fixtures/continue-go-microservice-039.okf.json
  • docs/reference/conformance/fixtures/deep-relation-graph-7-046.okf.json
  • docs/reference/conformance/fixtures/embedded-json-label-049.okf.json
  • docs/reference/conformance/fixtures/factory-droid-typescript-041.okf.json
  • docs/reference/conformance/fixtures/kiro-bash-ci-pipeline-040.okf.json
  • docs/reference/conformance/fixtures/large-entity-count-100-045.okf.json
  • docs/reference/conformance/fixtures/multi-modal-image-hint-050.okf.json
  • docs/reference/conformance/fixtures/opencode-python-debugger-038.okf.json
  • docs/reference/conformance/fixtures/rapid-fire-intent-stream-12-047.okf.json
  • docs/reference/conformance/fixtures/sql-migration-multi-intent-043.okf.json
  • docs/reference/conformance/fixtures/unicode-intent-label-cjk-048.okf.json
  • docs/reference/conformance/fixtures/yaml-k8s-deployment-044.okf.json
  • scripts/corpus-generate.py
  • tests/replay_breadth.rs
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch feat/sl-w44-b6-corpus
✨ Simplify code
  • Create PR with simplified code
  • Commit simplified code in branch feat/sl-w44-b6-corpus

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.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@gemini-code-assist

Copy link
Copy Markdown

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

KooshaPari pushed a commit that referenced this pull request Jul 24, 2026
Cherry-pick of 95a1049 to main. Updates WAVE44_SCOPE.md and docs/ops/WAVE44_PERT.md to:

- Mark R-5 (C01 L16 Fluent migration) as **withdrawn** — was a factual error (C01 L16 closed in Wave-38 #312)
- Update W44-B1 (loom-sse-soak) as SHIPPED via PR #372 — merge commit 774a927 already on main
- Update W44-B6 (corpus-breadth) as SHIPPED via PR #368 — pending merge
- Update realistic post-W44 to 397-399/402 (was 398-401)
- Resolve D-W44-3 (R-5 was a factual error, no machine lane needed)

Refs: PR #372 (W44-B1), PR #368 (W44-B6), WAVE44_SCOPE.md, docs/ops/WAVE44_PERT.md
@KooshaPari
KooshaPari merged commit 13c974f into main Jul 24, 2026
99 of 116 checks passed
@KooshaPari
KooshaPari deleted the feat/sl-w44-b6-corpus branch July 24, 2026 11:00
KooshaPari added a commit that referenced this pull request Jul 26, 2026
…3c974f (#374)

- SCORECARD.md: header refreshed (date=2026-07-24, auditor=Wave-44-D,
  commit=13c974f7); Wave-44 Delta section added (W44-B1 loom deepened C00 L7;
  W44-B6 corpus deepened C08 L73 20 -> 33 fixtures; PERT correction
  withdrawn R-5); Held (no score) line expanded with #368, #372, #373;
  Remaining unpaid rewritten to honestly name C04 L36 / C08 L76 /
  C11 L110 as the 3 unpaid pillars (6 raw pts).
- TRACEABILITY.json: updated 2026-07-23 -> 2026-07-24; commit 41829e8 ->
  13c974f; wave Wave-43 -> Wave-44; delta_vs_w42 -> delta_vs_w43. CRLF
  preserved.
- GAP_QA_MATRIX.md: C00 row mentions #372 loom HTTP SSE soak; C08 row
  mentions #368 corpus breadth and L76 residual; PLAN-W8-B row
  updated with Wave-44 closure refs and Wave-45 candidate lanes.
- CHANGELOG.md: Unreleased Changed entry for Wave-44-D reaudit.

Score 396/402 (98% A) held conservative; 2 of 3 machine lanes shipped
2026-07-24 (W44-B1 #372, W44-B6 #368). PERT correction #373. Remaining 6
raw pts across C04 L36 (HUMAN 2FA), C08 L76 (Agent-Eval Pipeline),
C11 L110 (Tray/Menubar Client) -- the 3 unpaid pillars the rubric
actually scores below 3.

Co-authored-by: KooshaPari <koosha@example.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant