[ARROW-16300] pc.sort_indices with nonexistent column throws malloc error

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 7.0.0
• Fix Version/s: 8.0.0
• Component/s: C++, Python
• Environment: OS X 11.6.1 MacBook Pro (13-inch, M1, 2020) ; Python 3.10.0 ; pyarrow 7.0.0
• External issue URL: https://github.com/apache/arrow/issues/31688
• Assignee: Unassigned
• Reporter: Ben Schmidt
• Created: 24/Apr/22 19:59
• Updated: 11/Jan/23 11:43
• Resolved: 26/Apr/22 03:13

Description

 

When calling pa.compute.sort_indices using a column name that is not in the table to be sorted, I get a malloc error that brings down the whole process, rather than a recoverable error.

Example:

import pyarrow as pa
from pyarrow import compute as pc
tb = pa.table({'foo': [1, 2, 3]})
pc.sort_indices(tb, [('foo', 'ascending'), ('bar', 'ascending')]) 

results in:

python(11346,0x1046cfd40) malloc: *** error for object 0x16bdae458: pointer being freed was not allocated
python(11346,0x1046cfd40) malloc: *** set a breakpoint in malloc_error_break to debug