Use chunks_exact for has_true/has_false to enable compiler unrolling

[adriangb]

Summary

• Replace .chunks(64) with .chunks_exact(16) in has_true() and has_false() as suggested in Add has_true() and has_false() to BooleanArray #9511 (comment)
• With chunks_exact, the compiler can fully unroll the inner fold (guaranteed size, no inner branch/loop), allowing a smaller block size for more frequent short-circuit exits without regressing the full-scan path

Benchmark results (block size 16 vs baseline)

• Full-scan worst case (65536): No regression (~49ns both)
• Early-exit cases (65536): ~27% faster (6.0ns → 4.4ns)
• Small arrays (64, 1024): Unchanged

Test plan

• All 13 existing test_has tests pass

run benchmarks boolean_array

@Dandandan Would appreciate your review!

🤖 Generated with Claude Code

[adriangb]

run benchmark boolean_array

[adriangb]

cc @Dandandan

[adriangbot]

🤖 Arrow criterion benchmark running (GKE) | trigger
Linux bench-c4080068357-397-rwfw5 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux
Comparing improve-has-true-false (e96e0e6) to d426107 (merge-base) diff
BENCH_NAME=boolean_array
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench boolean_array
BENCH_FILTER=
Results will be posted here when complete

[adriangbot]

🤖 Arrow criterion benchmark completed (GKE) | trigger

Details

group                                improve-has-true-false                 main
-----                                ----------------------                 ----
has_false(all_false, 1024)           1.00      7.8±0.03ns        ? ?/sec    1.08      8.5±0.09ns        ? ?/sec
has_false(all_false, 64)             1.00      8.2±0.08ns        ? ?/sec    1.00      8.1±0.09ns        ? ?/sec
has_false(all_false, 65536)          1.00      7.8±0.03ns        ? ?/sec    1.72     13.5±0.02ns        ? ?/sec
has_false(all_true, 1024)            1.00      7.8±0.05ns        ? ?/sec    1.11      8.7±0.16ns        ? ?/sec
has_false(all_true, 64)              1.00      8.3±0.14ns        ? ?/sec    1.02      8.5±0.07ns        ? ?/sec
has_false(all_true, 65536)           1.00     87.8±0.15ns        ? ?/sec    1.40    123.3±0.88ns        ? ?/sec
has_false(mixed_early, 1024)         1.00      8.2±0.04ns        ? ?/sec    1.07      8.7±0.12ns        ? ?/sec
has_false(mixed_early, 64)           1.00      8.3±0.08ns        ? ?/sec    1.03      8.5±0.12ns        ? ?/sec
has_false(mixed_early, 65536)        1.00      8.2±0.03ns        ? ?/sec    1.67     13.7±0.03ns        ? ?/sec
has_false(nulls_all_true, 1024)      1.02     19.2±0.06ns        ? ?/sec    1.00     18.7±0.14ns        ? ?/sec
has_false(nulls_all_true, 64)        1.15      7.7±0.04ns        ? ?/sec    1.00      6.7±0.06ns        ? ?/sec
has_false(nulls_all_true, 65536)     1.00    939.2±4.09ns        ? ?/sec    1.00    942.3±2.69ns        ? ?/sec
has_true(all_false, 1024)            1.00      7.7±0.05ns        ? ?/sec    1.09      8.4±0.03ns        ? ?/sec
has_true(all_false, 64)              1.01      8.2±0.09ns        ? ?/sec    1.00      8.2±0.08ns        ? ?/sec
has_true(all_false, 65536)           1.00     88.2±0.08ns        ? ?/sec    1.37    121.1±0.35ns        ? ?/sec
has_true(all_true, 1024)             1.00      7.5±0.03ns        ? ?/sec    1.16      8.6±0.02ns        ? ?/sec
has_true(all_true, 64)               1.01      7.9±0.07ns        ? ?/sec    1.00      7.8±0.07ns        ? ?/sec
has_true(all_true, 65536)            1.00      7.5±0.03ns        ? ?/sec    1.84     13.8±0.03ns        ? ?/sec
has_true(nulls_all_true, 1024)       1.03      6.5±0.11ns        ? ?/sec    1.00      6.3±0.10ns        ? ?/sec
has_true(nulls_all_true, 64)         1.04      6.5±0.11ns        ? ?/sec    1.00      6.3±0.10ns        ? ?/sec
has_true(nulls_all_true, 65536)      1.03      6.5±0.10ns        ? ?/sec    1.00      6.3±0.10ns        ? ?/sec
true_count(all_false, 1024)          1.01     10.0±0.04ns        ? ?/sec    1.00      9.9±0.02ns        ? ?/sec
true_count(all_false, 64)            1.00      8.7±0.34ns        ? ?/sec    1.01      8.8±0.30ns        ? ?/sec
true_count(all_false, 65536)         1.00    228.3±0.09ns        ? ?/sec    1.00    228.7±0.57ns        ? ?/sec
true_count(all_true, 1024)           1.00     10.0±0.03ns        ? ?/sec    1.02     10.2±0.02ns        ? ?/sec
true_count(all_true, 64)             1.00      8.7±0.31ns        ? ?/sec    1.05      9.2±0.29ns        ? ?/sec
true_count(all_true, 65536)          1.00    228.3±0.07ns        ? ?/sec    1.00    228.6±0.12ns        ? ?/sec
true_count(mixed_early, 1024)        1.01     10.3±0.03ns        ? ?/sec    1.00     10.2±0.02ns        ? ?/sec
true_count(mixed_early, 64)          1.00      9.1±0.32ns        ? ?/sec    1.02      9.2±0.30ns        ? ?/sec
true_count(mixed_early, 65536)       1.00    229.0±0.10ns        ? ?/sec    1.00    228.2±0.08ns        ? ?/sec
true_count(nulls_all_true, 1024)     1.03     18.9±0.03ns        ? ?/sec    1.00     18.4±0.03ns        ? ?/sec
true_count(nulls_all_true, 64)       1.02      7.3±0.06ns        ? ?/sec    1.00      7.2±0.19ns        ? ?/sec
true_count(nulls_all_true, 65536)    1.00    719.2±1.60ns        ? ?/sec    1.00    717.5±1.76ns        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	315.5s
Peak memory	1.9 GiB
Avg memory	1.9 GiB
CPU user	314.6s
CPU sys	0.7s
Disk read	0 B
Disk write	1.1 GiB

branch

Metric	Value
Wall time	318.2s
Peak memory	1.9 GiB
Avg memory	1.9 GiB
CPU user	318.0s
CPU sys	0.2s
Disk read	0 B
Disk write	2.8 MiB

[Dandandan]

Wooohoo! 🚀