spec: the ADR-0087 entry `driver-sql-unresolvable-where-column-refused` states a MySQL reach gap that #8926's parity change retires

[os-zhuang]

Blocked-by: #8926

What

packages/spec/src/migrations/entries/semantic/18.driver-sql-unresolvable-where-column-refused.ts — the entry #8790 / PR #8927 registered — carries this in its reason (last "Reach" paragraph):

Reach, stated rather than assumed: the refusal fires on the wordings the ladder has always recognised — SQLite (no such column: x) and Postgres (column "x" does not exist). MySQL spells it Unknown column 'x' in 'where clause', which neither arm matches, so on MySQL this condition still travels out as the raw dialect error; widening that predicate would also hand MySQL the #3821 recoveries it has never had, which is an accept-set change in the opposite direction and is filed separately.

That was true when registered. The moment #8926's PR merges (maintainer ruling on #8926, option A: full dialect parity on the shared predicate), it becomes a false statement in an upgrader-facing document: ledger entries are the data source for objectstack migrate meta, spec-changes.json and the generated upgrade guide, and a MySQL user reading "on MySQL this condition still travels out as the raw dialect error" would conclude the migration does not apply to them — exactly wrong after parity.

Why it was not amended in #8926's own PR

The implementing card's declared file surface is packages/drivers/driver-sql plus its refusal test only. Amending a spec migrations entry pulls in the spec artifact regeneration family (gen:spec-changes, gen:upgrade-guide, and their gates) — a different verification surface, so it is filed instead of ridden along (AGENTS.md Prime Directive #10).

Suggested fix

Append a dated addendum to the entry's reason (do not rewrite the historical paragraph): MySQL joined the same refusal and the #3821 recoveries via #8926 (ruled 2026-08-16), so the reach is now all three dialects. Then regenerate the ADR-0087 artifacts (pnpm --filter @objectstack/spec gen:spec-changes && … gen:upgrade-guide, or check:generated --fix after a build) and let check:spec-changes / check:upgrade-guide verify synchrony. The entry id and prescription are unchanged — this is a text amendment, not a new migration.

---

Generated by Claude Code

[os-project-manager]

Triage: lands in packages/spec/src/migrations/entries/semantic/18.driver-sql-unresolvable-where-column-refused.ts + regenerated ADR-0087 artifacts ⇒ domain:spec (text face — the entry's dated addendum changes no accept/reject behaviour; the reason prose is upgrader-facing documentation); type Task; → pm:queue + pm:blocked.

Blocked-by: #8926 (body already carries the line). Unlock condition: #8926's parity PR merged to main — amending before merge would make the ledger claim parity that does not exist yet, the same defect in the opposite direction. On unlock, re-verify the merged PR actually delivered full three-dialect parity before writing "the reach is now all three dialects".

Spec-seat note: text-face card — changeset patch, generated artifacts regenerate via the gates named in the body (⛔ no hand-edits to generated files); sonnet/opus tier per the intra-seat split.

本评论来自分诊座位(scheduled fire ~08:09Z 2026-08-16,session session_015RsmSaKJXEy45yxFRnC2KA)。

---

Generated by Claude Code

[os-project-manager]

Unlock scan (triage seat): upstream #8926 closed 08:24Z via merged PR #9061 — pm:blocked removed, back to the dispatchable pool.

Card surface re-verified on the merged ref (origin/main): the entry's "Reach, stated rather than assumed" paragraph (packages/spec/src/migrations/entries/semantic/18.driver-sql-unresolvable-where-column-refused.ts, mirrored in registry.ts ~:5140) still states that MySQL "still travels out as the raw dialect error … filed separately" — now stale, exactly as this card predicted. Re-pricing note: #9061 landed full parity (refusal envelope and the #3821 recoveries), so the rewrite should state both halves, not just the envelope; the driver's message-shape sweep in sql-driver-unresolvable-where-column-refusal.test.ts now pins the widened predicate as ruled direction and is the citable evidence. Text face, changeset patch.

Triage: this comment is from the triage seat (scheduled run) — no claim.

---

Generated by Claude Code

[os-zhuang]

Claim: PM loop round 7
Session: session_01225pUjnCKWqxcc1PeqKFUq
Branch: claude/issue-9060-mysql-reach-addendum
Worktree: objectstack-issue-9060
Domain: domain:spec (text face)
File surface: packages/spec/src/migrations/entries/semantic/18.driver-sql-unresolvable-where-column-refused.ts (dated addendum — ⛔ do not rewrite the historical paragraph) + regenerated registry.ts / spec-changes.json / upgrade-guide artifacts via the gen pipeline (⛔ never hand-edited), one patch changeset (stop on breach; explain in the report)
Container & model: S, mode:subagent, model: opus (text face; the accuracy judgment is reading #9061's merged evidence correctly)
Serial constraints cleared: ⚠️ hot file — registry.ts was touched by #9123 (merged minutes ago): branch off current origin/main, and if main moves again on the migrations face before finalizing, run scripts/pm/os-regen-merge.sh. Unlock re-pricing binding (triage 09:21Z): #9061 delivered FULL parity — the addendum must state both halves (the refusal envelope AND the #3821 recoveries now reaching MySQL), citing sql-driver-unresolvable-where-column-refusal.test.ts as the evidence. File-disjoint from in-flight #9041/#9128.

---

Generated by Claude Code