[ARROW-12521] [C++] arrow-compute-scalar-set-lookup-benchmark failure with new random generator

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Component/s: Benchmarking, C++
• External issue URL: https://github.com/apache/arrow/issues/28285
• Assignee: Unassigned
• Reporter: Antoine Pitrou
• Created: 23/Apr/21 17:04
• Updated: 11/Jan/23 08:26
• Resolved: 27/Apr/21 14:31

Description

Unfortunately, the value set lookup kernels don't support duplicate values in the value set, which makes our benchmark for these kernels inherently fragile.

-- Arrow Fatal Error --
NotImplemented: duplicate values in value_set

subprocess.CalledProcessError: Command '['/tmp/arrow-archery-du8luj93/WORKSPACE/build/release/arrow-compute-scalar-set-lookup-benchmark', '--benchmark_repetitions=1', '--benchmark_out=/tmp/tmpn20z4ul9', '--benchmark_out_format=json']' died with <Signals.SIGABRT: 6>.

Issue Links

• is superceded by: ARROW-12554 Allow duplicates in the value_set for compute::is_in