Treat OpenAPI path parameters as required in v4 import This tracks unresolved feedback…
- User
Treat OpenAPI path parameters as required in v4 import This tracks unresolved feedback identified by cubic on PR #1149: https://github.com/withcoral/coral/pull/1149#discussion_r3347190640
Context: This appears to be pre-existing DSL v4 behavior surfaced by the refactor review, not a regression from the module split.
Problem: OpenAPI path parameters are imported as optional when
requiredis omitted, even though path parameters are required by OpenAPI. This can misclassify required path inputs as optional SQL filters.Acceptance criteria: Import path parameters as required regardless of an omitted
requiredfield, preserve explicit behavior for non-path parameters, and add focused importer/projection coverage.Interface
No new interfaces are introduced.
- Assistant
I’ll trace the v4 OpenAPI importer and its projection tests, then patch the requiredness rule and run focused tests plus the mandated Rust checks.
- Assistant
rgisn’t installed in this checkout, so I’m falling back to targetedfind/grepsearches. - Assistant
The bug is the importer’s blanket
required.unwrap_or(false). I’ll make path location authoritative (path || explicit required) and add one IR-level test plus one projection-level test that proves the SQL surface becomes a required function argument instead of an optional filter. - Assistant
Both focused tests pass. I’m checking artifact-version conventions now, because changing import semantics may require an importer version bump so previously materialized v4 packages fail loudly and can be re-added, consistent with this repo’s v4 rules.
- Assistant
The full
coral-specsuite passes (221 unit tests plus doctests). I’m running the repository’s requiredmake rust-checksgate next. - Assistant
The gate caught only test-style lint violations from direct indexing. I’m replacing those with explicit
first().expect(...)bindings, then I’ll rerun the full gate. - Assistant
Formatting and workspace Clippy now pass. The gate’s test phase cannot start because
cargo-nextestis not installed locally; I’m running the equivalent workspace/all-targets/all-features suite with Cargo’s built-in test runner instead. - Assistant
The fallback workspace suite has reached integration tests; two unrelated source-lifecycle rollback tests have failed while the rest continue. I’ll capture their failure details and rerun them individually to distinguish an environment/concurrency issue from a real failure.
- Assistant
The two failures reproduce individually and are caused by this container running as root: both tests expect Unix
0500permissions to block writes/deletes, which root bypasses. They are unrelated to this change. I’m doing one final package test run after formatting and then handing off with that environment limitation documented.
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.
Identical first prompt
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- fable-5Compare side by side →
- fable-5Compare side by side →