join: optimize locale collation performance with hybrid comparison

[sylvestre]

Performance improvements:

• 5.18x faster than upstream locale collation
• 35% faster than GNU join for Unicode data
• Maintains full locale collation correctness

[github-actions]

GNU testsuite comparison:

Skipping an intermittent issue tests/tail/follow-name (passes in this run but fails in the 'main' branch)

[ChrisDryden]

Were you expecting the results of this change to show up here? #10391

[ChrisDryden]

I think this is not a valid assumption:

Heres a one liner that shows that for punctuation byte order does not always equal collation order. Its part of the reason the test was changed in this pr I think

echo -e "ab:d 1\nabc:d 2" | LC_ALL=en_US.UTF-8 join --check-order - <(echo -e "ab:d x\nabc:d y")

[sylvestre]

oh, well spotted

[sylvestre]

Were you expecting the results of this change to show up here? #10391

yeah, i am confused why it isn't detected

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/pr/bounded-memory is no longer failing!

[github-actions]

Binary size comparison:

Individual binary size comparison VS main (threshold: >=5% AND >=4 KB).

Total size of compared binaries: 148.57 MB (+972 KB, +0.64%)

Significant per-binary changes:
  ls         2.11 MB ->    3.25 MB  (+1.14 MB, +54.36%)
  mkfifo     1.16 MB ->    1.08 MB  (-76 KB, -6.42%)
  mknod      1.16 MB ->    1.09 MB  (-76 KB, -6.38%)
  mkdir      1.16 MB ->    1.09 MB  (-72 KB, -6.04%)

[codspeed-hq]

Merging this PR will degrade performance by 4.76%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 2 regressed benchmarks
✅ 345 untouched benchmarks
⏩ 46 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	ls_recursive_deep_tree[(200, 2)]	3.4 ms	3.7 ms	-6.13%
❌	Simulation	du_summarize_balanced_tree[(5, 4, 10)]	15.5 ms	16 ms	-3.38%

Tip

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

---

_{Comparing sylvestre:join-perf (472116d) with main (7cc5cfe)²}

Footnotes

1. 46 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. No successful run was found on main (bfe212f) during the generation of this report, so 7cc5cfe was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩