fix(lint): descend into conditional then/otherwise when building the _validations universe

[baozhoutao]

Fixes #14700

What

validate-translation-references built its _validations universe with a flat walk of objects[].validations[]. A conditional rule's then / otherwise branch is itself a full rule carrying its own name, and that branch name — not the wrapper's — is the address checkConditional delegates to and authoredRuleMessage keys on at runtime (packages/objectql/src/validation/rule-validator.ts). The flat walk never saw a branch name, so a correct bundle entry for one was reported as an orphan translation-target-unknown, and the finding's own advice ("keeps its source locale in every refusal") was the opposite of the truth for that key — acting on it (deleting the entry) reintroduces the exact defect it fixed.

Fix

buildUniverse's walk over obj.validations now calls a small recursive collector (collectValidationRuleNames) instead of reading rule.name directly:

function collectValidationRuleNames(rule: AnyRec, validations: Set<string>): void {
  const ruleName = strName(rule.name);
  if (ruleName) validations.add(ruleName);
  if (isRec(rule.then)) collectValidationRuleNames(rule.then, validations);
  if (isRec(rule.otherwise)) collectValidationRuleNames(rule.otherwise, validations);
}

This mirrors evaluateRule's recursion (checkConditional dispatches the branch back through evaluateRule, which recurses into checkConditional again for a nested conditional) — depth is unbounded, matching the runtime.

The wrapper's own name stays in the universe, unchanged. Its message is structurally unreachable (checkConditional either returns nothing, returns unevaluableRuleError — which builds its own sentence — or delegates to the branch), so an entry for it is inert at runtime. #14518 keeps one anyway, deliberately, so the bundle mirrors the declared rule set 1:1 rather than re-deriving objectql's dispatch — this PR does not touch that decision.

packages/objectql/** is untouched: checkConditional / authoredRuleMessage are the addressing this mirrors, not changes.

Before / after (reproduces the card's own measurement)

Same fixture as the issue — demo_account with one conditional rule (churn_reason_consistency) whose then / otherwise are churn_reason_present / churn_reason_absent, and a bundle with entries for all three names:

Tree	findings.length	Findings
origin/main (e6ac0c6fd5, pre-fix)	2	translation-target-unknown on churn_reason_present and churn_reason_absent
this branch (post-fix)	0	—

Reproduced by importing validateTranslationReferences directly (via tsx) from a comparison worktree checked out at the pre-fix commit and from this branch, against the identical fixture object — not just asserted in a test, run both ways.

Fixtures added (validate-translation-references.test.ts, new describe block #14700)

Case	Bundle names	Expected findings
Wrapper + both branches named	churn_reason_consistency, churn_reason_present, churn_reason_absent	0 (was 2 before the fix)
A name matching no rule at any depth	the two real branch names + churn_reason_ghost	1 — churn_reason_ghost only (real orphans still caught)
Nested-in-nested (conditional whose then is itself a conditional)	innermost_rule (2 levels deep)	0
Unnamed branch	gate (wrapper only; branch has no name)	0, no crash (mirrors the existing unnamed-top-level-rule behaviour, #14253)

All 64 tests in the file pass, including the pre-existing object-branch coverage vs the schema (#13835) pin, which still asserts one working leg per ObjectTranslationDataSchema key including _validations.

Gates

• pnpm --filter @objectstack/lint test — 93 test files / 2845 passed, 5 skipped (pre-existing skips, unrelated).
• pnpm --filter @objectstack/lint typecheck — clean (tsc --noEmit + check:test-typecheck).
• eslint --no-inline-config on both touched files — 0 errors / 0 warnings.
• node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --ran <record> at 413167ed8c (post-merge-of-main) — 34 derived, 30 run (all green), 4 NOT-MEASURED, 0 UNRUN:
  • check-test-completeness — grades a saved turbo run test log; none exists locally (CI tees one).
  • check-half-states (plain, live sweep) — timed out at 180s reaching GitHub through the session proxy; report-only, never fails a build (its own header says so); the check:pm-half-states self-test alias ran and passed.
  • check:dual-build-cjs-loads — reads built dist/ across ~80 workspace packages; full pnpm build not run in this worktree.
  • check:type-check-debt — --re-measure refuses without the full package build closure, by design (a number measured without it would count a different world, per its own message).

None of the 4 relate to this diff's file surface (packages/lint/src/validate-translation-references.{ts,test.ts}); all are repo-wide gates whose local prerequisite is a full monorepo build/test run, which CI performs.

Changeset

.changeset/nested-conditional-validation-refs.md — @objectstack/lint patch.

---

🤖 Generated with Claude Code

https://claude.ai/code/session_01WLJQhde67SeTccsmnBVarV

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

3 anchor(s) derived from 1 changed package(s); no hand-written page names any of them, so this run has nothing to list — not a clean bill of health. This check sees only pages that NAME a derived anchor: one that documents this change in prose, or enumerates it in an authoring dialect, names none and stays invisible to it on every run.

What this run could not see

• the SDK route bridge reached 47 of 219 client-bound route-ledger rows — the other 172 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 172: 14 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 56 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 102 are undecided (no in-repo declaration, on a ledger that has other in-repo registrars — absence and an unreadable spelling are not distinguishable here). The rows themselves: node scripts/docs-audit/affected-docs.mjs --bridge-coverage
• 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 — 5 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 2d40f9146800dcf625fc54696f6435e676b1ed96 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 28e6348288a4174df92fb64f5fc8409ac0fcd2c6 — the merge of head c2095e53852f6702f3625c348f92067f254e1282 into base 2d40f9146800dcf625fc54696f6435e676b1ed96, which is what actions/checkout gives a pull_request run. Not the PR head.

A worktree cut from an older main holds a different content/docs, so re-deriving there can legitimately return a different list — that is a different tree, not a wrong row. To answer on the same tree:

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin 28e6348288a4174df92fb64f5fc8409ac0fcd2c6 && git checkout 28e6348288a4174df92fb64f5fc8409ac0fcd2c6
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 2d40f9146800dcf625fc54696f6435e676b1ed96 c2095e53852f6702f3625c348f92067f254e1282 && git checkout -B drift-repro 2d40f9146800dcf625fc54696f6435e676b1ed96 && git merge --no-ff c2095e53852f6702f3625c348f92067f254e1282

node scripts/docs-audit/affected-docs.mjs --json 2d40f9146800dcf625fc54696f6435e676b1ed96

⚠️ That checkout carried uncommitted changes, so the commit above does not fully identify what was read.

[claude]

Re-trigger via update-branch (domain:devx execution seat, session session_01LAwHpn4uVuf4N1geBcD5i3).

PR-side CI on 413167ed8 is red only on Test Core (1/6) (packages/cli pnpm run test exit 1 — check-run annotations carry the exit code, the log download is refused from this container; same shard/package as the #14648 signature, and this diff does not touch packages/cli test code). This head does not contain the merged fix accb9231c7 (PR #14715; verified git merge-base --is-ancestor), so this is the "base lacks a merged fix ⇒ merge main, push a new head" branch, not a re-run: GitHub merges main into the branch (a real merge commit, no rewrite). If Test Core (1/6) reds again on the new head with the same file, it parks under the #14648 post-fix reading like PR #14774; any other red is real. #14700 stays pm:dispatched; flip + arm on green.

---

Generated by Claude Code