feat(ir): name un-aliased aggregate columns by source text (#598/#599) — corpus 555→570

[DecisionNerd]

What

An aggregate RETURN item with no AS — RETURN count(*), min(x), sum(i) — was named agg_<idx>, so it failed the TCK's header-exact result comparison (column_by_name("count(*)")). lower_return_aggregate now uses the item's verbatim expression text (ReturnItem::display, the same source #906 introduced for non-aggregate columns), falling back to agg_<idx> only when neither an explicit alias nor display text exists.

This is the dominant Aggregation-corpus pattern (single top-level aggregate, no AS), which the binder already lowered correctly — only the output column name was wrong.

Impact

Corpus passing 555 → 570 (+15, 0 regressions); baseline re-blessed. No golden churn (the IR aggregation golden uses an aliased aggregate).

Follow-up

Naming the group-by keys of a mixed aggregate query (RETURN n.name, sum(n.num) → the n.name column is still named by its lowered expr, not n.name) needs per-key output aliases on the Aggregate IR op — a separate PR. Nested aggregates (count(*) + 1) need the Aggregate→Project decomposition, also separate.

Verification

$ cargo test --workspace --no-fail-fast   # ALL GREEN
$ cargo test -p gf-api --test bdd         # 570 passing, 0 regressed
$ cargo clippy --workspace -- -D warnings # clean

Refs #598/#599 (does not close).

🤖 Generated with Claude Code

Note

Name un-aliased aggregate RETURN columns by their source expression text

• In binder.rs, aggregate RETURN items without an explicit alias now use the item's verbatim display text (e.g., count(*), min(x)) as the column name, matching openCypher naming behavior, with fallback to agg_<n> only when no display text exists.
• TCK passing scenarios increase from 555 to 570, with new passing entries across Aggregation, CountingSubgraphMatches, Match, Return, and Comparison feature files.

^{Macroscope summarized 2ed78ac.}

Summary by CodeRabbit

• Bug Fixes

  • Improved default column naming for aggregate results so returned values use clearer, more consistent labels when no explicit alias is provided.
  • This brings aggregate result naming in line with other returned expressions, reducing unexpected generic column names.

• Tests

  • Expanded query compliance coverage with additional passing scenarios for aggregation, comparisons, graph match counting, deprecated pattern matching, and return-clause aliasing/interoperation.
  • Increased total passing scenarios in the compatibility matrix.

[coderabbitai]

Walkthrough

In lower_return_aggregate, the alias resolution for un-aliased aggregate RETURN items is updated to fall back to item.display before defaulting to the synthetic agg_<idx> name. The TCK passing baseline gains 15 new scenario entries across several feature groups, and the coverage matrix scenarios_passing count is updated from 555 to 570.

Changes

Aggregate alias fallback and TCK coverage update

Layer / File(s)	Summary
Aggregate alias fallback logic crates/gf-ir/src/binder.rs	lower_return_aggregate now resolves un-aliased items as item.alias → item.display → agg_<idx> instead of the previous two-way item.alias → agg_<idx>.
TCK baseline and coverage matrix tests/tck/passing_baseline.txt, tests/tck/coverage_matrix.json	Adds 15 new passing scenario lines covering Aggregation*, Comparison1, CountingSubgraphMatches1, Match9, Return4, Return5, and Return8; bumps scenarios_passing from 555 to 570.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• DecisionNerd/graphforge#906: Updates lower_return_items in binder.rs to use ReturnItem.display as the fallback for un-aliased non-aggregate RETURN projections — the parallel change to the one in this PR for the aggregate path.
• DecisionNerd/graphforge#891: Also modifies tests/tck/coverage_matrix.json's scenarios_passing metadata count in the same field updated here.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly states the IR change and the corpus impact, matching the main changeset.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
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.
Description check	✅ Passed	The description clearly explains the change, impact, verification, and follow-up work, so it is sufficiently complete despite not mirroring the full template.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/599-aggregation-completeness

---

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