183

[RAprogramm]

Closes #183.

Real schemas carry constraints the macro cannot infer — foreign keys over natural keys, custom-named CHECK constraints, indexes from hand-written migrations. Found while validating typed_constraints against a real backend, where FK violations still required hand-written constraint-name matching.

constraint(name, kind, field?)

• Repeatable inside #[entity(...)]; kinds: unique, foreign_key, check; field optional
• Entries merge into the generated registry and take precedence over auto-derived entries with the same name (single match arm, custom field wins)
• Validation: requires typed_constraints; missing name/kind and unknown kinds/options rejected with clear errors

Testing

• 4 registry/parse unit tests (merge, precedence/dedupe, gating, kind validation) — 672 total green
• Extra gating test pinning that sql::generate emits nothing without the postgres feature (covers the branch that slipped Codecov in 184 #188)
• trybuild pass case extended with a custom FK declaration
• clippy -D warnings clean, all-features suite green, deny ok — checklist run before commit

Docs

• README typed-constraints section extended; wiki (Attributes ×5) updated in this PR via wiki/

[codecov]

Codecov Report

❌ Patch coverage is 99.19355% with 1 line in your changes missing coverage. Please review.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...ity-derive-impl/src/entity/parse/entity/helpers.rs	98.11%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!