Allow every object type as a dependency key (L8)

[gmr]

Closes the last substantive item from the review-findings audit (L8).

Problem

ObjectType::from_plural_key (src/constants.rs) already recognizes all 25 object types, but schemata/dependencies.yml listed only nine plural keys under additionalProperties: false. Because per-object validation runs before the dependencies block is stripped, a dependencies: block naming an aggregate, collation, event_trigger, materialized_view, publication, server, subscription, user, or user_mapping was rejected at validation — you could depend on views but not materialized_views. Carried over from Python's OBJ_KEYS; per the maintainer, 2.0 need not preserve that limitation.

Change

• Add the nine missing keys to schemata/dependencies.yml (schema-qualified schema.name pattern for the schema-scoped types; bare pattern for the schemaless ones), alphabetized with the existing keys.
• Add a validate test asserting the widened keys validate and unknown keys are still rejected by additionalProperties: false.

Resolution already works end-to-end for these types: the loader's index_key normalizes the namespace to None for schemaless types on both insert and lookup, and from_plural_key maps the keys — so no loader change is needed. (propertyNames on a type: array is a JSON-Schema no-op, matching the existing entries, so the patterns are cosmetic; additionalProperties is the functional gate.)

Validation

• just check — fmt, clippy -D warnings, all tests pass (new dependencies_schema_accepts_every_object_type).
• just gates — round-trip + all three deploy gates pass.

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features

  • Expanded support for more dependency categories, including aggregates, collations, materialized views, publications, servers, subscriptions, user mappings, and users.

• Tests

  • Added coverage to confirm the schema accepts the broader set of dependency object types.
  • Verified unknown dependency types are still rejected.

[coderabbitai]

Important

Review skipped

No new commits to review since the last review.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: f1d6fae9-ed16-472e-bef7-c9c6d9ba1664

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

📝 Walkthrough

Walkthrough

This PR expands the JSON Schema for PostgreSQL dependency objects to include new top-level properties (aggregates, collations, materialized_views, publications, servers, subscriptions, user_mappings, users) and adds a unit test validating the widened schema and rejection of unknown keys.

Changes

Dependencies schema expansion

Layer / File(s)	Summary
Schema property definitions schemata/dependencies.yml	Adds new array-backed properties for aggregates, collations, materialized_views, publications, servers, subscriptions, user_mappings, and users, and repositions domains, event_triggers, and sequences, each with regex-constrained propertyNames.
Schema validation test src/project/validate.rs	Adds a unit test that validates a dependency object using all supported plural keys with no errors, and confirms an unknown key is still rejected under additionalProperties: false.

Estimated code review effort: 2 (Simple) | ~10 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: expanding dependency keys to allow all object types.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/dependency-schema-keys

---

_{Comment @coderabbitai help to get the list of available commands.}

[gmr]

🤖 This comment was posted by Claude on behalf of @gmr

CI is green and there are no open review threads, but the CodeRabbit status check is still pending. Re-triggering a review.

@coderabbitai review

[coderabbitai]

@gmr Got it, kicking off a fresh review now.

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

src/project/validate.rs (1)

129-159: 🎯 Functional Correctness | 🔵 Trivial | ⚡ Quick win

Test correctly exercises validator_for/iter_errors, but silently doesn't catch the propertyNames-on-array schema gap.

The test data uses arrays of plain strings for every widened key (e.g. "aggregates": ["test.agg"]), which is exactly the instance shape for which propertyNames in dependencies.yml is a no-op. As a result this test only verifies that the keys are accepted (which they would be even with malformed names like "aggregates": ["not-a-schema-qualified-name"]) — it does not actually verify name-format enforcement. Once the propertyNames/items issue in schemata/dependencies.yml is fixed, consider adding a negative case here with a malformed item value (e.g. "aggregates": ["missing_dot"]) to confirm the pattern constraint is enforced.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/project/validate.rs` around lines 129 - 159, The dependencies validation
test in validate::dependencies_schema_accepts_every_object_type only checks that
widened keys pass validator_for/iter_errors, but it never verifies item-name
pattern enforcement because the schema gap is on array items rather than the
keys themselves. Update this test to add a negative case using a malformed
dependency name value for one of the widened keys, and assert
validator.iter_errors returns an error so the pattern constraint is actually
exercised alongside the existing accepted/unknown-key checks.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@schemata/dependencies.yml`:
- Around line 7-22: Update the dependency array schemas in the dependencies
definition so the string validation applies to array elements instead of object
keys; in the affected dependency entries like aggregates, collations, domains,
and event_triggers, replace propertyNames-based validation with items-based
pattern validation, and make the same change for the other dependency arrays in
this file so each dependency string is actually checked.

---

Nitpick comments:
In `@src/project/validate.rs`:
- Around line 129-159: The dependencies validation test in
validate::dependencies_schema_accepts_every_object_type only checks that widened
keys pass validator_for/iter_errors, but it never verifies item-name pattern
enforcement because the schema gap is on array items rather than the keys
themselves. Update this test to add a negative case using a malformed dependency
name value for one of the widened keys, and assert validator.iter_errors returns
an error so the pattern constraint is actually exercised alongside the existing
accepted/unknown-key checks.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: f9b6b7da-6744-445c-9966-7697f5da12a5

📥 Commits

Reviewing files that changed from the base of the PR and between 31d3485 and af752b0.

📒 Files selected for processing (2)

• schemata/dependencies.yml
• src/project/validate.rs

[gmr]

PR monitoring complete — ready to merge.

• CI: green (test suite success)
• Review threads: 0 unresolved. CodeRabbit's propertyNames suggestion was declined with rationale and the thread resolved — the propertyNames patterns on type: array are a JSON-Schema no-op that this PR merely extends from the pre-existing entries; additionalProperties: false is the functional gate, and rewriting them into items.pattern would reject valid array contents.
• Reviewers: CodeRabbit reviewed current HEAD (af752b0).

No code changes were needed this session beyond addressing review feedback.