[ARROW-12911] [Python] Export scalar aggregate options to pc.sum (sum of zero rows gives null; should give 0)

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 4.0.1
• Fix Version/s: 5.0.0
• Component/s: Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/28639
• Assignee: Yibo Cai
• Reporter: Adam Hooper
• Created: 31/May/21 14:58
• Updated: 11/Jan/23 08:29
• Resolved: 04/Jun/21 10:15

Description

>>> pa.compute.sum(pa.array([], pa.int64()))
<pyarrow.Int64Scalar: None>

I'd expect 0.

I can't think of any reason for NULL, except that SQL returns NULL. But I can't figure out why SQL returns NULL. Does anybody know? Any textbook – and https://en.wikipedia.org/wiki/Summation – specifies 0.

Pandas and Numpy return 0. Also, Apache Arrow c_glib implementation returns 0 – and even tests for it: https://github.com/apache/arrow/blob/master/c_glib/test/test-int8-array.rb#L60

Workaround is to replace all NULLs with 0 after running the computation.

Issue Links

• links to: GitHub Pull Request #10433