test(driver-sql): budget the #13056 live-MySQL orphan-shadow tests

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.

[zhuangjianguo]

Fixes#13688

The defect

Temporal Conformance (live PG + MySQL) reddens unrelated PRs — it surfaced on
#13685, whose entire diff is packages/plugins/plugin-auth/** + a changeset + one
docs page, touching no driver, no SQL, no schema-sync path.

The two it(...) blocks in
sql-driver-13056-orphan-shadow-column-cleanup.test.ts that read the live MySQL
catalog each drive 2-3 full new SqlDriver(...) -> initObjects(...) ->
disconnect() cycles against the live container, with no explicit timeout, so they
inherited vitest's 5000ms default. Both observed failures are timeouts, not
assertion failures, and the service-container logs show no MySQL error — the DDL
that ran succeeded. This is not "raise a timeout to make a slow test pass": the
5000ms was never a chosen budget for 2-3 live connect/DDL/disconnect cycles plus
several information_schema reads — it's what a test gets by default when nobody
sets one. Four sibling files in this same package already carry explicit per-test
budgets (60_000 / 120_000 / 40_000 / 60_000); this file shipped with #13056's
fix without one.

The cost of this defect is not local — it reddens other people's PRs on a
package they never touched, so every timeout here is a repo-wide throughput tax
(ejected merge-queue attempts, wasted CI minutes), not a local annoyance. See #13767,
the automated queue-flake anchor for this exact file, ejected twice in 24h.

The fix

Both it(...) blocks get an explicit 60_000 budget — in line with 3 of the 4
sibling budgets in this package — with a one-line comment naming why (N live
round trips against MySQL), so the next reader does not delete it as noise.

git diff on packages/drivers/driver-sql/src/*.ts excluding this one test file is
empty — no production code, no assertion, and no control changed.

A2.1 — the acceptance control (actual cost against live MySQL)

NOT MEASURED. No live MySQL/PG container is reachable in this environment (the
Docker daemon is not running here — docker info fails with no such file or directory on the socket; no OS_TEST_MYSQL_URL/OS_TEST_POSTGRES_URL set; no
mysql/service listening on 3306/5432). Leaning on the structural argument instead:
timeout (not an assertion failure), no MySQL error in the logs, and four sibling
files in the same package already carrying comparable budgets for a comparable
number of live round trips.

A2.2 — sweep for the same shape elsewhere

Confirmed: only these two it(...) blocks in this file lacked a budget (both
now fixed). A package-wide sweep for the same shape (any it(...) nested in a
declareDialectCell(...) callback, still on vitest's inherited 5000ms default) found
9 more files, 37 more it(...) blocks with the identical gap. Per the dispatch
order's STOP condition 3 ("a package-wide re-budget is a different, larger
decision"), the fix here stays scoped to this one file; the sweep result is filed
as #13902 for a deliberate decision rather than folded into this PR.

A2.3 — no global testTimeout

Confirmed by reading packages/drivers/driver-sql/vitest.config.ts: no
testTimeout key anywhere in the test block. The two it(...) blocks really were
falling through to vitest's own default, not a repo-set global.

A2.4 — anchors re-located by quoted source

The card cites sql-driver-13056-orphan-shadow-column-cleanup.test.ts:499. Re-found
both it(...) blocks by their quoted titles rather than trusting the line number: the
first ("drops the orphaned generated column, and keeps the one still carrying a
constraint") is still at line 499 — no drift. The second ("converges — a second
detect finds neither the index nor an orphan column") is at line 555.

Tests

• pnpm --filter @objectstack/driver-sql typecheck — PASS (exit 0), via
  scripts/pm/os-verify-lock.sh.
• pnpm --filter @objectstack/driver-sql exec vitest run --maxWorkers=2 — PASS:
  147 passed | 9 skipped test files (156), 2252 passed | 132 skipped tests (2384).
  The target file alone (--reporter=verbose, single-file run): 12 passed | 1
  skipped — the whole live-MySQL describe block collapses to ONE named-skip
  test ("is provisioned — set OS_TEST_MYSQL_URL to run this cell"), so the two
  edited it(...) blocks did not individually execute here. All non-live cells
  run against embedded SQLite and execute fully. The MySQL/PG cells this PR
  touches cannot run here — no live MySQL/PG container in this environment
  (see A2.1), so declareDialectCell reports them as a named SKIP rather than a
  silent pass (see the file's own header comment on OS_TEST_MYSQL_URL /
  OS_EXPECT_LIVE_DIALECT_MATRIX). They will run for real in the Temporal Conformance (live PG + MySQL) CI job this PR exists to stop reddening.
• Local gate family (node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack,
  re-derived at HEAD fcc9ce8034 after git fetch origin main): 26 relevant
  families (22 path-matched + 4 more from the "edits a test file" convention
  trigger) — 23 PASS, 3 NOT MEASURED:
  • check:dual-build-cjs-loads, check:type-check-debt — need the FULL
    workspace built (pnpm build / turbo run build across all 78 packages),
    which is CI's own "Build Core" / debt-ledger scope, not warranted locally
    for a one-file test-only diff (driver-sql's own dependency closure was
    built and its tsc --noEmit passes clean).
  • check:test-completeness — by its own design, exits PREREQUISITE NOT MET
    without a saved turbo run test log; CI tees that log and passes it, this
    branch is unreachable there.
    None of the 23 that ran flagged anything in this diff. The dispatch-gates
    derivation itself carried a STALE-TREE warning (its own defining scripts had
    moved on origin/main by the time of the final re-run, in a repo with several
    other agents landing PRs concurrently) — re-running after git fetch origin main reproduced the identical 22-family path-matched list, so nothing was
    missed by the staleness.

No ablation

No ablation is meaningful here — there is no behaviour to neuter. The change is a
test-timing budget, not logic; there is nothing for a mutation to disprove that a
rebuild-and-rerun would show.

Changeset

skip-changeset — this PR's entire diff is one test file (a timeout budget +
explanatory comments); it publishes nothing from any package.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json 0c143ececbd5aa92c6edbf1d84bbf82668b585c9 → packageMentionDocs.