[connortsui20]

Summary

The DecimalByteParts cast rule pushed every nullability cast into its most significant part (MSP), including casts to non-nullable that it could not prove safe. The cast-rule contract on Validity::cast_nullability says that a CastReduce rule, which runs in the optimizer without execution, must decline when Validity::trivially_cast_nullability cannot decide, and that a CastKernel handles the cast at execution. No wrong result is visible today, because the null check still runs when the MSP cast executes, but the rule relied on every MSP encoding keeping that check.

Changes

The rule now declines when trivially_cast_nullability returns None. A new CastKernel checks the validity with cast_nullability and then rebuilds the array over the cast MSP, reusing the lower parts. Casts that change the precision or the scale stay unsupported.

When the MSP has a validity array, a cast to non-nullable now stays a lazy Cast over the DecimalByteParts array until execution, instead of becoming a DecimalByteParts array in the optimizer. The MSP's own cast kernel checks the validity a second time, which for a primitive MSP is a lookup of the cached minimum.

The cast tests now use a session that registers the crate, so they reach the new kernel. Seven other encodings push nullability casts the same way, tracked in #10750.

🤖 Generated with Claude Code

[codspeed]

Merging this PR will degrade performance by 1.68%

4 warnings, 1 note

⚠️ 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

⚠️ Unable to measure the execution of 2 benchmarks

These benchmarks measured no execution time, so their results are not comparable and count as unchanged. This usually happens when the compiler removes the code under test.

Prevent compiler optimizations

See affected benchmarks

⚠️ 21 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚠️ Base and head ran on different runner types

Some benchmarks with significant performance changes were compared across different runner types, so the changes might not come from the code. Run base and head on the same runner type, or use CodSpeed Macro Runners for more stable results.

Learn more about Macro Runners

Open the report in CodSpeed to investigate

ℹ️ 40 benchmarks have no flamegraph

The flamegraph wasn't generated: the symbols of the profiling trace could not be resolved, because no information about the language frames was collected.

See affected benchmarks

⚡ 3 improved benchmarks
❌ 6 regressed benchmarks
✅ 2399 untouched benchmarks
⏩ 12 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	CPU Simulation	test_scan_filter_to_arrow_array[int_float]	3.5 ms	4.2 ms	-16.39%
❌	Walltime	starts_with_avx512[8, VarBin]	39.5 µs	45.8 µs	-13.62%
❌	Walltime	starts_with_avx2[64, VarBin]	54.1 µs	62.6 µs	-13.58%
❌	CPU Simulation	test_repeated_scan_filter[int_float]	3.4 ms	3.9 ms	-12.89%
❌	CPU Simulation	test_repeated_scan_filter[int]	2.4 ms	2.7 ms	-10.22%
❌	CPU Simulation	test_scan_filter_to_arrow_array[int]	2.5 ms	2.8 ms	-10.11%
⚡	CPU Simulation	bench_compare_primitive[(10000, 32)]	97 µs	72.9 µs	+33.09%
⚡	CPU Simulation	chunked_dict_primitive_canonical_into[f64, (1000, 1000, 10)]	191.7 µs	151.6 µs	+26.44%
⚡	Walltime	starts_with_avx2[8, VarBin]	46.9 µs	40.3 µs	+16.29%

Tip

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

---

_{Comparing ct/dbp-nullability-cast (87589b7) with develop (3974ee0)}

Footnotes

1. 12 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. ↩