[ARROW-10437] [C++][Dataset] Ensure virtual columns are not materialized on write

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.0.0
• Component/s: C++
• Labels: dataset
• External issue URL: https://github.com/apache/arrow/issues/26415
• Assignee: Unassigned
• Reporter: Ben Kietzman
• Created: 30/Oct/20 20:26
• Updated: 11/Jan/23 08:13

Description

If a column is virtual on read and will remain virtual after writing then materializing and grouping on it is a pure waste. Maybe Partitioning::Partition should take a "already known" partition expression and only group on columns not in it?