Skip to content

docs: close-out CC-203 / CC-247 / CC-248 — harness adoption complete (22/23)#153

Merged
screenleon merged 1 commit into
mainfrom
docs/cc-247-248-203-closeout
May 24, 2026
Merged

docs: close-out CC-203 / CC-247 / CC-248 — harness adoption complete (22/23)#153
screenleon merged 1 commit into
mainfrom
docs/cc-247-248-203-closeout

Conversation

@screenleon
Copy link
Copy Markdown
Owner

Summary

Docs-only close-out for 3 stale-active reuse-debt tickets. Code shipped in earlier PRs (#127 / #128 / #135-#140 / #142 / #152) but BACKLOG status was never flipped.

Each ticket gets an **Outcome** + **See** block in its body section.

Validator impact

30 → 29 E-codes (BACKLOG.md). Opportunistically cleared E-AREA-ENUM: CC-203 invalid area: ops/test/reuseops/test while touching the row (per [[validator-dryrun-before-strengthen]] pattern: data prep tagging along with the close-out edit). Other 29 baseline errors unchanged.

Test plan

  • bash pm/scripts/validate.sh BACKLOG.md DECISIONS.md CHANGELOG.md — 29 E-codes (down from 30 baseline, 1 cleared)
  • No code touched — git diff --stat origin/main..HEAD shows BACKLOG.md only
  • Verified grep -lE '^(pass|fail)\(\)' scripts/test-*.sh returns zero (CC-247/248 acceptance)
  • Verified 22 of 23 scripts/test-*.sh use th_init (CC-203 acceptance) — sole non-adopter is the orchestrator
  • Verified 90a600a feat(cc-247,cc-248): th_init --format=<preset> + --fail-fast harness options (#142) provides both options

Gate skipped per [[gate-architecture-not-data]] — pure BACKLOG.md data commit, validator-neutral (improving), no code path touched.

🤖 Generated with Claude Code

…e (22/23) + format/fail-fast options + consumer adoption

CC-247 + CC-248 (th_init --format=<preset> + --fail-fast harness options)
shipped in PR #142 on 2026-05-23 but BACKLOG status was never flipped.
Final consumer adoption verified post PR-B.2 v2 (#152) close-out:
- 4 of 10 migrated consumers select a non-default --format preset
- 3 of 4 fail-fast consumers use --fail-fast (test-skill-refine,
  test-usage-tracker, test-usage-weekly)
- grep '^(pass|fail)()' scripts/test-*.sh returns 0 (no per-file
  print-format or fail() overrides remain)

CC-203 (shared scripts/lib/test-harness.sh) reaches its closure
threshold at 22/23 test-*.sh adoption. The lone non-adopter is the
orchestrator scripts/test-run-all-tests.sh which runs other test-*.sh
as subprocesses and asserts on aggregated output — not a case-based
harness fit; deliberately excluded per the GROUP-B re-analysis. The
3 files excluded from PR-B.2 v2's assert_* migration (CC-256) are
on a separate axis and do not block CC-203 closure.

Index + body sections updated for all 3 tickets with Outcome / See
blocks. Validator-neutral edit (30 → 29 E-codes — opportunistically
cleared CC-203 invalid area enum `ops/test/reuse` → `ops/test` as
part of the row touch).
@screenleon screenleon merged commit 6b3e740 into main May 24, 2026
16 checks passed
screenleon added a commit that referenced this pull request May 24, 2026
…err noise gone) (#154)

* fix(cc-257): escape backticks in pr-gate.sh codex-brief heredocs (stderr noise gone)

Every /pr-gate invocation emitted 7 `final::` / `**Final::` / `Final::`
command-not-found errors at scripts/pr-gate.sh:362 (BRIEF_EOF heredoc
opening line). Same 3-line block also lived in the synthesis brief
heredoc (SBRIEF_P2). Root cause: unquoted heredoc delimiter → bash
performed command substitution on the 7 backticked `Final:`/`final:`
tokens that CC-252 (#147) added as cautionary examples ("DO NOT emit
`**Final: GO**` etc.").

Fix: escape the 7 backtick pairs (` → \`) in 3 unique lines that appear
in both BRIEF_EOF (lines 451/452/467) and SBRIEF_P2 (lines 839/840/860).
Bash now writes the backticks literally; brief content unchanged from
codex's perspective — the cautionary tokens still render as backticked
code in markdown.

Considered Option A (quote the heredoc delimiter `<< 'BRIEF_EOF'`) but
rejected: would have required pre-substituting 12 `${VAR}` references
+ `$(date)` — a much larger refactor for the same functional outcome.
Option B (escape backticks) is a 6-line surgical edit.

Regression test added at scripts/test-pr-gate.sh::
test_brief_construction_emits_no_shell_errors — invokes run_gate with
CODEX_GATE_CAPTURE_BRIEF set, asserts stderr file does NOT contain
"command not found" AND the captured brief still carries the cautionary
tokens (verifies fix without losing the warning).

Local verification:
- test-pr-gate.sh 48/48 (was 47, +1 from new regression case)
- test-test-harness.sh 30/30
- lint-scripts 52 files OK
- Validator-neutral (29 → 29 E-codes; CC-203 area fix already in
  origin/main via PR #153)

CC-257 row + body section flipped to ✅ closed 2026-05-24 with
Outcome + See blocks.

* docs(cc-257): tail commit — update pr:TBD → pr:#154
@screenleon screenleon deleted the docs/cc-247-248-203-closeout branch May 24, 2026 13:46
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