[connortsui20]

Summary

• Stacked on Execute interleaved decimal arrays #10940.

An out-of-bounds take index panicked or returned an error depending on the encoding. Fixed-width take (and so Dict execution) panics, while RunEnd take, Chunked take and Interleave execution returned an error. An index past the end is a contract violation by the caller, not invalid input, so all of them now panic.

Changes

Interleave cannot be serialized, so its selectors always come from code in the same process. Its bool and fixed-width kernels now panic on a selector that names a missing value or row, which also matches Interleave::scalar_at. RunEnd and Chunked take panic with the same message as the fixed-width kernel.

Take indices can still come from a file, for example the codes of a dict layout. A corrupt file with an out-of-bounds code now panics for RunEnd and Chunked values, as it already does for primitive values.

Follow-up: FixedSizeList, Null and Filter take, and the fixed-width take of a PiecewiseSequence, still return an error.

API Changes

take on a RunEnd or Chunked array, and execution of an Interleave array, panic on an out-of-bounds index where they used to return an error.

[connortsui20]

Closing for now. It is not decided whether an out-of-bounds take index must panic or return an error. The branch ct/interleave-take-panic stays for reference.

[codspeed]

Merging this PR will degrade performance by 16.32%

3 warnings

⚠️ 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 benchmark is 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

⚡ 3 improved benchmarks
❌ 12 regressed benchmarks
✅ 2259 untouched benchmarks
⏩ 370 skipped benchmarks¹
🗄️ 30 archived benchmarks run²

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	CPU Simulation	filter_powerlaw_by_random[10000]	670 ns	1,901 ns	-64.76%
❌	Walltime	mul_u64_nonnull_neon	20.1 µs	39.2 µs	-48.82%
❌	CPU Simulation	bench_compare_varbin[(50000, 2)]	133.8 µs	162.2 µs	-17.49%
❌	CPU Simulation	bench_compare_varbin[(50000, 64)]	135.6 µs	163.8 µs	-17.25%
❌	CPU Simulation	vortex[round_robin/n2/nonnull]	246.7 µs	283.7 µs	-13.04%
❌	CPU Simulation	vortex[random/n2/nonnull]	247.5 µs	284.4 µs	-12.99%
❌	CPU Simulation	vortex[round_robin/n2/null]	251.2 µs	288.1 µs	-12.8%
❌	CPU Simulation	vortex[random/n2/null]	252.2 µs	289.1 µs	-12.76%
❌	CPU Simulation	bench_compare_sliced_dict_varbinview[(7500, 10000)]	200.8 µs	227.7 µs	-11.81%
❌	CPU Simulation	bench_compare_sliced_dict_varbinview[(10000, 10000)]	207 µs	234 µs	-11.53%
❌	CPU Simulation	bench_compare_sliced_dict_varbinview[(1000, 10000)]	79.9 µs	89 µs	-10.19%
❌	CPU Simulation	interleave[canonical]	1.2 ms	1.3 ms	-10.01%
⚡	CPU Simulation	random_i8[0.01]	123.3 µs	98.7 µs	+24.95%
⚡	CPU Simulation	max_f64	315.2 µs	282.1 µs	+11.73%
⚡	Walltime	compare_u64_avx2	3.6 µs	3.3 µs	+10.67%

Tip

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

---

_{Comparing ct/interleave-take-panic (b0c7c8f) with ct/decimal-interleave (c0eb004)}

Footnotes

1. 370 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. ↩

2. 30 benchmarks were run, but are now archived. If they were deleted in another branch, consider rebasing to remove them from the report. Instead if they were added back, click here to restore them. ↩