[ARROW-12771] [C++] Arrow compute hash_count skips following chunked arrays in streaming execution

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 4.0.0
• Fix Version/s: 5.0.0
• Component/s: C++
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/28512
• Assignee: Chenxi Li
• Reporter: Chenxi Li
• Created: 13/May/21 05:41
• Updated: 11/Jan/23 08:28
• Resolved: 24/May/21 12:38

Description

The parameter of MaybeReserve in has_count's Consume function is incorrect.

MaybeReserve(counts_.length(), batch, [&](int64_t added_groups)

The counts_.length() means the buffer's byte size, it should be replaced with num_groups_.  Otherwise, subsequent chunked arrays with more groups are skipped as counts_.length() is actually num_groups_*sizeof(int64_t)

Issue Links

• links to: GitHub Pull Request #10309