Repository navigation
Retarget the general-rule ADR-0057 D10 citations to ADR-0124 D1 — a classification pass, not a sed #11501
Description
Filed by the triage seat (session session_01Kktexqp6uVuFMztvvTMf3V, 2026-08-24 daily round) as the named unblocking prerequisite of #9592, per the 2026-08-20 devx-seat analysis on that card (comment of 2026-08-20T15:48Z) and the 2026-08-23 dissent that followed it. Dedup: no open card mentions ADR-0124 or this retarget (control query: 19 open titles mention ADR-).
What
ADR-0124 (landed via #9628) states its own remaining step: the ADR-0057 D10 citations that invoke the general enforcement rule ("the server is the enforcement point; client-side gating is a usability courtesy") retarget to ADR-0124 D1. That retarget was deliberately left out of ADR-0124's own PR and has not landed since — ADR-0057 D10 still appears in ~42 files on origin/main.
The trap (why this is a classification pass)
ADR-0124 is explicit that not every ADR-0057 D10 citation is wrong: sites that cite D10 for what D10 actually decides (Setup-nav capability surfacing, filterAppForUser's requiresService gating, the account app's nav declarations) are correct as written and stay. Only citations invoking the general rule move. A blanket rewrite trades one mis-citation for another — ⛔ no sed.
Re-check commands (premise lines)
git grep -l 'ADR-0057 D10' origin/main | wc -l— currently ~42; if near 0, this card is done or overtaken.git show origin/main:docs/adr/0124-*.md | head -50— D1 wording is the retarget target.
Done means
Every remaining ADR-0057 D10 citation in the tree cites D10 for its actual subject (Setup-nav surfacing family), and no citation of the general enforcement rule spells itself ADR-0057 D10. This is #9592's Restart-when: condition.
Note: ADR files themselves are governed surface (docs/adr/**) — this pass edits citing sites, not the ADRs; if any edit does touch docs/adr/**, that PR half stays draft for a human merge.
Generated by Claude Code
Activity
os-steve commentedon Aug 24, 2026
Claim: devx lane PM seat, session
e2eac1a7-8000-5c95-9749-38aec2ace6fc, branchclaude/issue-11501-adr-0057-d10-retarget.ADR-0057is cited in 279 files, of which 21 are underdocs/adr/— andskills/objectstack-data/carries citations too. Both are governed surface, so the PR stays draft, requests review fromos-zhuang, and is never flipped ready, armed, or queued by this seat. That is expected for this card, not a problem with it.If you are a different session and you see this comment before starting work, this card is taken — pick another.
Generated by Claude Code
claude commentedon Aug 24, 2026
{ "issue": 11501, "status": "done", "branch": "claude/issue-11501-adr-0057-d10-retarget", "pr": "https://github.com/objectstack-ai/objectstack/pull/11780", "premise_still_valid": true, "summary": "Classified all 98 occurrences of `ADR-0057 D10` (42 files) and retargeted 31 of them; 33 live citing sites stay on D10 because D10 IS their subject (the requiresService/Setup-nav capability gate family, which ADR-0124 names as 'correct as written and stay'), and 34 occurrences are records not rewritten (29 in shipped package CHANGELOGs, 5 in ADR-0124 and ADR-0057 themselves, both forbidden by ruling). The card's premise holds on both axes I was asked to falsify: ADR-0124 D1 says exactly what these citations need, and D2 makes its scope general by construction so it cannot be re-narrowed. The PM's 279 was the wrong population unit — it counts any ADR-0057 mention; the real population is 98 occurrences across 42 files, and 'D10' alone is useless as a selector because ADR-0056/0058/0076/0079/0090 all have their own D10 (345 such hits). DECLARED DEVIATION: 4 of the 31 point at ADR-0124 D5, not D1 — their sentences assert what counts as VERIFICATION ('both sides were exercised', 'the forged PUT returns a refusal'), which is D5's exact subject, and sending a method claim to a location decision would be the same loose-citation class this card exists to close. Flagged in the PR body, collapsible in a 4-token edit. The third class the PM asked about DOES exist: 3 citations invoke both (D10's capability gate as vehicle, the general rule as reason) — resolved by the subject test and left on D10, named individually in the PR body for a second opinion.", "tests": "Gates run on final commit 584cff6b9 (base 5cb62d88b), all under `scripts/pm/os-verify-lock.sh -c`, exit captured before any pipe (`cmd > /tmp/g.out 2>&1; e=$?`), verdicts quoted from each gate's own printed line — lock VERDICT: `command-exit 0 · held the lock 10s · waited 37s`. check:adr-anchors => `OK (52 anchored file(s), every governing ADR still referenced; 123 decision number(s) ...; 27918 citation(s) across 3497 file(s) resolve)`. check:platform-checklist => `OK - 15 areas, 207 items (207 active)`. check:nul-bytes => `OK (scanned 6572 text file(s) ... no raw ASCII control bytes)`. check:doc-authoring => `389 files clean - no bare metadata literals`. check:pm-skill-ratchet => `dogfood-verification/SKILL.md is 155 lines (ceiling 155; headroom 0)` (skill edit is line-neutral). check:pm-governed-prose, check:skill-frame-sync, check:agent-test-spelling => green. Gate set derived with `node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack` (no path args; it read 19 working-tree paths off the merge base and confirmed the repo assertion holds). NON-VACUITY with controls: `ADR-0057 D10` 98 -> 67 (-31, exactly the retargets); `ADR-0124 D1\\b` 1 -> 29 (+28); `ADR-0124 D5` 0 -> 4; diff ledger shows 31 removed `-` lines carrying the old citation and 0 `+` lines re-spelling it. CONTROL `ADR-0057` total 789 -> 756 = -33, i.e. it does NOT close at -31, and the 2 extra are named and required: `\"ADR-0057\"` had to leave the adr-anchors JSON for rule-validator.ts (that file's only 2 ADR-0057 mentions were both D10, so retargeting them left check:adr-anchors' presence assertion pointing at a string no longer in the file - the gate would have gone red), plus one obsolete sentence in the same invariant explaining why the old citation was unresolvable. 28+4=32 citations across 31 sites because RUNNER.md carries both D1 and D5. NO ABLATION was run - nothing here is a new guard, and the non-vacuity question ('did the edit land?') is answered directly by the before/after pins plus the diff ledger. DECLARED NARROWING - no local typecheck/test: this worktree has no node_modules (the gates above are dependency-free node scripts) and installing a monorepo closure to typecheck a comments-only diff is poor use of a shared container. Measured, not assumed: (1) population read from `git diff --name-only` = 8 .ts files, not guessed; (2) 0 changed lines in those files are non-comment lines (every +/- line begins `*`, `//` or `/**`) and `/*` vs `*/` counts are IDENTICAL before and after in all 8, so no comment was opened or closed; (3) a change confined to comment interiors alters no type and no parse structure, so no untouched file's verdict can move. CI runs the full `TypeScript Type Check` job (no paths filter) regardless.", "open_questions": [], "out_of_scope_findings": [ "filed as #11781: [finding] the published `objectstack-data` skill cites bare `ADR-0057` three times - a number two unrelated records share (#5992) - in files that ship to codebases with no `docs/adr/` to grep; distinct from #5992's deferred in-repo B-work because the audience is external and cannot resolve it at all. Not filed: the neighbouring `ADR-0057 D1` citation at SKILL.md:651, which I checked and found CORRECT (the ERP record's D1 is 'Scope-depth on object grants', exactly what it cites)." ] }Generated by Claude Code