Skip to content

docs: refresh §9 refactor state — R5 complete (ADR-0006 shipped) - #99

Merged
KraHsu merged 1 commit into
devfrom
docs/r5-refactor-state
May 22, 2026
Merged

KraHsu merged 1 commit into
devfrom
docs/r5-refactor-state

Conversation

@KraHsu

@KraHsu KraHsu commented May 22, 2026

Copy link
Copy Markdown
Owner

What

Refresh ROADMAP §9 to reflect Phase R5 (task-specific rewards split) complete — R5.1 (#97), R5.2 (#98) merged; ADR-0006 shipped. Docs-only PR, matching #93 / #96.

Changes (ROADMAP.md)

  • §9.0 status: phase header → R0–R5 complete; status table R5 row → ✅ complete (PRs refactor: mdp/motion_tracking.py owns the motion-imitation rewards (ROADMAP §9 PR R5.1 / ADR-0006) #97/refactor: motion_body_error_exp factory dedups the motion-tracking trio (ROADMAP §9 PR R5.2 / ADR-0006) #98); ADR-0006 → Accepted; R7 row regated on "R6 landing + cross-layer follow-ups".
  • Completed + Tests-run entries for R5.1 (whole-family verbatim move; the ADR-named-3 vs actual-6 variance) and R5.2 (the motion_body_error_exp factory; jaccard-1.000 trio only; bit-equivalence test; 395 → 398 tests).
  • R5.2 risk row marked resolved (the parameterization-drift risk was guarded by torch.equal against pinned originals).
  • §9.2 sequencing diagram + prose: R5 ✅ — only R6 remains before R7.
  • Phase R5 detail block: header → COMPLETE, scope/PRs/tests "as shipped", and completion criteria — including an honest note that the planned Unitree G1 reference-run check was superseded by the bit-equivalence test (no reference-runs doc exists yet; deferred to M1.7).
  • New lessons: (1) an ADR's enumerated symbol list goes stale — re-audit the module before a "move" slice; (2) for a merging refactor, pin the pre-refactor bodies in a test (no in-tree original to diff post-merge); (3) prefer thin def wrappers over functools.partial for back-compat shims (__name__/repr).

Note

CLAUDE.md and plans/ (incl. ADR-0006's Accepted flip + implementation outcome) are gitignored local artifacts; refreshed in the same pass for future local sessions, but only the tracked ROADMAP.md is committed here — same as #93 / #96.

🤖 Generated with Claude Code

Update ROADMAP §9 to reflect Phase R5 (task-specific rewards split) complete:
R5.1 (#97), R5.2 (#98) merged.

- §9.0 phase header + status table: R0–R5 complete; ADR-0006 → Accepted;
  R7 row regated on "R6 landing + cross-layer follow-ups".
- Add R5.1 / R5.2 "Completed" entries and "Tests run" lines (incl. the
  whole-family move variance and the bit-equivalence test).
- R5.2 risk row marked resolved.
- "Next steps" / "Suggested next slice" rewritten: R6 is the last structural
  R-phase; R5/R6 no longer "fan out" (R5 done).
- §9.2 sequencing diagram + prose: R5 ✅.
- Phase R5 detail block: header → COMPLETE, scope/PR/test "as shipped",
  completion criteria (jaccard collapse by construction; the Unitree G1
  reference-run check superseded by the bit-equivalence test — no
  reference-runs doc exists yet, deferred to M1.7).
- New lessons: an ADR's enumerated symbol list goes stale (re-audit before a
  move); pin pre-refactor bodies in a test for a *merging* refactor; prefer
  thin def wrappers over functools.partial for back-compat shims.

(CLAUDE.md and plans/ are gitignored local artifacts and were refreshed locally
in the same pass; only the tracked ROADMAP.md is committed, matching docs PRs
#93 and #96.)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@KraHsu
KraHsu merged commit c392d49 into dev May 22, 2026
3 checks passed
@KraHsu
KraHsu deleted the docs/r5-refactor-state branch May 22, 2026 07:39
KraHsu added a commit that referenced this pull request Jun 5, 2026
docs: refresh §9 refactor state — R5 complete (ADR-0006 shipped)
KraHsu added a commit that referenced this pull request Jun 5, 2026
…0009 shipped)

The architecture refactor is done: importlinter is now a required CI gate with a
clean 6-contract / 0-violation baseline. Update ROADMAP §9 accordingly.

- §9.0 phase header + status table: R6 (#100) and R7 (#101–#106) marked complete;
  ADR-0007/0008/0009 → Accepted; ADR-0010 deferred. Reviewed against dev @ b9f7e0a.
- Add "Completed" entries for R6, R7.1, R7.2, R7.3a–d and matching "Tests run"
  lines (incl. the R6 cv2/Qt subprocess-isolation incident and the R7.3d 6/0
  baseline).
- Dependency-graph note: 3,982 nodes / 11,433 edges post-R7 (net growth from new
  modules + deprecation shims; cross-layer imports 24 → 0).
- R4/R5.2/R7 risk rows marked resolved; the "21 violations gating R7" notes
  rewritten to reflect the cleared baseline.
- §9.2 diagram + prose: all phases ✅; Phase R6/R7 detail headers → COMPLETE.
- New lessons: layers-contract forbids intra-band siblings (use forbidden
  contracts); fix-the-layer beats DI; precise contract beats blanket ban + waivers;
  subprocess-isolate cv2-importing tests.
- "Next steps" rewritten: refactor complete; remaining items are optional
  follow-ups (BaseTermManager for metrics/curriculum is the recommended next
  slice; ADR flips; ≤400 cli target; test_cli split; shim removal; ADR-0010).

(CLAUDE.md, plans/adr/0007–0009, and the ADR status flips are gitignored local
artifacts, refreshed in the same pass; only ROADMAP.md is committed — matching
docs PRs #93 / #96 / #99.)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.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