refactor(tests): generalise scalar matrix harness off integer-inherent consts

[tobyhede]

Summary

PR 1 of 2 (stacked). Makes the SQLx scalar-matrix harness type-agnostic ahead of the first non-integer scalar, without adding one — so the date type (#256, stacked on this branch) becomes a clean catalog-row-plus-wiring change. Behaviour-preserving for the existing int4 / int2 types.

What changed

Three integer assumptions in the harness are lifted:

• ScalarType::FIXTURE_VALUES (a const) → fn fixture_values(). A method, so a scalar whose values can't be const-constructed (e.g. a chrono type) can return a borrow of a lazily-built Vec. Integer impls still hand back their eql_scalars::<T>_VALUES const.
• New min_pivot() / max_pivot() trait methods replace the matrix's direct <T>::MIN / <T>::MAX pivot references, so a scalar without an inherent ::MIN/::MAX const can supply an explicit sentinel.
• The ORDER BY arms build their WHERE clause from to_sql_literal(zero) instead of a hardcoded > 0, so a non-integer plaintext column typechecks.

The macro emits the new method shape for integer impls; the int4 cargo-expand snapshot is regenerated to track the method-based bodies.

Why behaviour-preserving

For int4 / int2: min_pivot/max_pivot resolve to Self::MIN/Self::MAX, to_sql_literal(0) renders 0 (so WHERE plaintext > 0 is byte-identical), and the generated test names are unchanged — the matrix_tests.txt inventory snapshot does not move.

Verification

• cargo test -p eql-tests-macros — green
• cargo check -p eql_tests --tests — clean (the const→fn rename compiles across all ~20 call sites)
• cargo fmt --check, clippy -D warnings — clean

Full PG matrix runs in CI. PR 2 (eql_v3.date) stacks on this branch.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Chores
  • Updated internal test infrastructure to improve fixture value and pivot handling within the test suite.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 1bdac2ad-096f-49d3-93f0-8da4c3305ffe

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:

• ✅ Review completed - (🔄 Check again to review again)

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch v3-scalar-harness-generalize

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[coderabbitai]

Actionable comments posted: 2

🤖 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 `@crates/eql-tests-macros/src/lib.rs`:
- Around line 268-273: The test currently only checks for the substrings
"MIN"/"MAX" which may appear in doc comments; update the assertions to verify
the actual emitted pivot function bodies for min_pivot and max_pivot instead of
loose substrings. Specifically, in the test that inspects out, replace the
out.contains("MIN") and out.contains("MAX") checks with assertions that match a
more precise snippet from the generated functions (e.g., the signature and the
body that returns the inherent bounds for min_pivot and max_pivot) so you
validate the functions min_pivot and max_pivot actually return the expected
bounds; keep the existing checks for "fn fixture_values" and the presence of the
pivot function names.

In `@tests/sqlx/src/scalar_domains.rs`:
- Around line 37-46: The contract for fixture_values() is ambiguous about
ordering but callers (e.g., ordered_numeric_matrix! driven code) depend on a
stable, insertion-consistent ordering and perform positional indexing; update
the trait/docs and implementations so fixture_values() guarantees a
deterministic order matching the fixture insertion order (or explicitly sorted
by a defined key) — modify the fixture_values() implementations for non-integer
scalars to build and return a statically stable slice (e.g., by producing a Vec
once in a deterministic order and leaking or storing it as &'static [Self]) and
update the trait comment for fixture_values() to state that callers may rely on
its ordering (or alternatively have callers sort before positional access if
ordering cannot be guaranteed).

🪄 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: defaults

Review profile: CHILL

Plan: Pro

Run ID: 39aa6b97-4a18-4a3e-8315-f083e656290e

📥 Commits

Reviewing files that changed from the base of the PR and between b45cea6 and b265e43.

📒 Files selected for processing (5)

• crates/eql-tests-macros/src/lib.rs
• tests/sqlx/snapshots/int4_expanded.rs
• tests/sqlx/src/matrix.rs
• tests/sqlx/src/scalar_domains.rs
• tests/sqlx/tests/encrypted_domain/family/mutations.rs

[tobyhede]

Fixes Applied Successfully

Fixed 2 file(s) based on 2 CodeRabbit feedback item(s).

Files modified:

• tests/sqlx/src/scalar_domains.rs — documented the fixture_values() stable-order contract (callers compare the slice element-wise against the ORDER BY id plaintext column and index positionally [0]/[len/2] without sorting; a lazily-built Vec must preserve fixture insertion order).
• crates/eql-tests-macros/src/lib.rs — assert the emitted min_pivot/max_pivot function bodies instead of loose MIN/MAX substrings that also match the doc comment.

Commit: 09d1de1

The latest autofix changes are on the v3-scalar-harness-generalize branch.