[connortsui20]

Summary

• Tracking Issue: Tracking Issue: Fix Extension Types #11114

ExtensionStoragePushdown returned the results of its aggregate over storage as they were, so vortex.min_max[vortex.timestamp] gave raw i64 values without the time unit or time zone. Plain min_max returns timestamps for the same column, and nothing restores the type later, so the pushdown now returns extension values too.

Changes

Through a new StorageResultMapping trait, MinMax and IsSorted name their value fields, the fields of their partials and results that hold input values. They name them by index, since a dtype can't tell them apart: over Bool storage, the is_sorted field also has the storage dtype. Grouped finalize receives partials whose value fields already hold extension values, so the trait also requires that finalize only moves or drops those fields and ignores args.dtype, as both aggregates already do.

The fixed mapping is a stopgap until #11115 gives every aggregate that runs on storage a result mapping through ExtensionUse::Storage. The impls sit next to the aggregates, so #10769 can move them along.

Tests compare the pushdown with plain min_max over flat, chunked, constant, dictionary-encoded, and grouped input, and check that an is_sorted partial with a null value survives its round trip through a scalar.

API Changes

ExtensionStoragePushdown<A, E> now requires A: StorageResultMapping, which only MinMax and IsSorted implement. The value fields of the results and partials, such as min and max of vortex.min_max[vortex.timestamp], now have the extension dtype, and a storage value that the extension type rejects is an error, as with plain min_max.

[codspeed]

Merging this PR will degrade performance by 0.81%

1 warning

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 1763 untouched benchmarks
⏩ 611 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	CPU Simulation	random_i8[0.01]	99.6 µs	113.3 µs	-12.09%
⚡	CPU Simulation	take[(16, 8, 1)]	120.8 µs	107.9 µs	+11.92%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ct/storage-pushdown-results (8d58672) with develop (bc58587)}

Footnotes

1. 611 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[connortsui20]

Closing in favor of #11243. Naming the value fields by index is a half-measure, and the claims work in #11115 replaces the adapter anyway, so the bug gets fixed there. The branch stays up, since its tests compare the pushdown with plain min_max and are useful for SP4.