fix(tesseract): Honor memberToAlias for granularized time dimensions

[ovr]

In a wrapped SQL-API query, cubesql pushes a grouped CubeScan whose granularized time-dimension column it references by an explicit alias, sent via memberToAlias keyed {member}.{granularity} (e.g. orders.created_at.month -> datetrunc_utf8__). Regular dimensions and measures already honor this override, but TimeDimensionSymbol computed its alias as {base alias}_{granularity} (orders__created_at_month) and ignored the override. The wrapper then referenced a column the CubeScan never emitted -> column "datetrunc_utf8__" does not exist.

Look up the memberToAlias override for the granularized member in compile_time_dimensions and pass it through a new TimeDimensionSymbol::new_with_alias. When no override is present the alias is unchanged, so non-pushdown queries are unaffected.

[claude]

Claude finished @ovr's task in 1m 8s —— View job

---

Code Review — Verdict: LGTM (1 low, 1 nit)

Review details

• Read diff for time_dimension_symbol.rs, query_properties_compiler.rs, common_sql_generation.rs
• Verify alias_for_member lookup key matches cubesql memberToAlias convention
• Check derived-symbol code paths (change_granularity, reference_member, get_dependencies_as_time_dimensions)
• Test coverage

Summary

The fix is well-scoped and correct: when compile_time_dimensions builds a TimeDimensionSymbol, it now consults evaluator_compiler.alias_for_member("{member}.{granularity}") and threads the result through a new TimeDimensionSymbol::new_with_alias. When no override is present, the alias falls back to {base alias}_{granularity}, so non-pushdown paths are untouched. Comments explain the dotted-key convention from cubesql.

Observations

Low — derived-symbol constructors still call Self::new and drop the override
time_dimension_symbol.rs:119,172,257 (change_granularity, get_dependencies_as_time_dimensions, reference_member) all construct a fresh TimeDimensionSymbol via Self::new, which forwards alias_override = None. For change_granularity this is arguably correct — the override is keyed by the old granularity, so a re-granularized symbol shouldn't inherit it. For reference_member and get_dependencies_as_time_dimensions the behavior is murkier: a wrapped-reference variant of the same logical member loses the explicit projection alias. If the SQL-API wrapper ever flows through one of those paths the original bug could resurface. Worth a follow-up to either (a) propagate the override or (b) document why these paths are guaranteed not to reach the wrapper.

Nit — alias_override = Some("") would silently win over the default
time_dimension_symbol.rs:53-54 uses unwrap_or_else, so an empty-string override would produce an empty SQL alias. cubesql shouldn't send that today, but a .filter(|s| !s.is_empty()) on the lookup site would make this self-defensive.

Praise

• The doc-comments on the call site and on new_with_alias clearly capture why the dotted key is used and what the fallback is — exactly the kind of WHY that survives refactors.
• The test (test_member_to_alias_time_dimension_granularity) asserts both the positive (override appears) and negative (default alias is absent) cases — that pairing is what catches a regression where both aliases get emitted.

• [branch](https://github.com/cube-js/cube/tree/fix/tesseract-member-to-alias-granularitzed-td)

[claude]

Claude finished @ovr's task in 1m 2s —— View job

---

Code Review

Verdict: LGTM — small, well-scoped fix. No issues found.

Details

• Read PR diff (4 files, +121/-7)
• Verify implementation against described root cause
• Check test coverage (Rust unit + JS integration)
• Check for dedup vs existing review threads (none)

What the fix does

TimeDimensionSymbol previously hard-coded its alias to {base alias}_{granularity}, ignoring memberToAlias. The fix:

• Adds TimeDimensionSymbol::new_with_alias(...) accepting an Option<String> override, with the old new(...) delegating to it (None). Backward-compatible — non-pushdown queries unaffected.
• In compile_time_dimensions, looks up memberToAlias for the key {base.full_name}.{granularity} (the dotted key cubesql actually sends) and threads it through. Lookup only runs when d.granularity.is_some(), which matches the wrapper's behavior.

Correctness notes

• query_properties_compiler.rs:253-259: the dotted-key format matches the cubesql side that emits memberToAlias for granularized columns — confirmed against the PR description.
• time_dimension_symbol.rs:53-54: when override is absent, falls back to format!("{}_{}", base_symbol.alias(), name_suffix) — identical to prior behavior. The base symbol's alias itself already honors its own memberToAlias override, so the un-granularized case is untouched.
• name_suffix defaults to "day" when granularity is None; the override lookup is gated on granularity.is_some(), so a user keying memberToAlias as member.day without an explicit granularity wouldn't be honored. That's fine — cubesql sends an explicit granularity when it sends the dotted alias key — but worth being aware of.

Tests

• common_sql_generation.rs: unit test asserts both that the override appears ("td_month_alias") and that the default fallback does not (visitors__created_at_month). Good positive + negative assertion.
• sql-generation.test.ts: end-to-end integration test exercises the full SQL API path with multiple memberToAlias entries (regular dim + measures + granularized td) and checks result-row column names. Good coverage.

Nits / non-blocking

• The // Honor an explicit memberToAlias override … comment block in query_properties_compiler.rs:249-252 is a touch heavy on per-CLAUDE.md "comments explain WHY not WHAT" — but in this case the why (dotted-key contract with cubesql) is genuinely non-obvious to a future reader, so it earns its place. Fine as-is.
• No changelog entry needed — Lerna will handle it on release.

• Branch: `fix/tesseract-member-to-alias-granularitzed-td`

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 79.18%. Comparing base (4d4b8cd) to head (47b7cc3).

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #11142   +/-   ##
=======================================
  Coverage   79.18%   79.18%           
=======================================
  Files         474      474           
  Lines       96880    96880           
  Branches     3558     3558           
=======================================
+ Hits        76712    76714    +2     
+ Misses      19650    19648    -2     
  Partials      518      518           

Flag	Coverage Δ
cube-backend	58.75% <ø> (+0.01%)	⬆️
cubesql	83.68% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.