[SPARK-42750] Support INSERT INTO by name

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.4.0
• Fix Version/s: 3.5.0
• Component/s: SQL
• Labels: pull-request-available
• Assignee: Jia Fan
• Reporter: Jose Torres
• Created: 10/Mar/23 17:13
• Updated: 10/Sep/23 00:18
• Resolved: 07/Jun/23 02:31

Description

In some use cases, users have incoming dataframes with fixed column names which might differ from the canonical order. Currently there's no way to handle this easily through the INSERT INTO API - the user has to make sure the columns are in the right order as they would when inserting a tuple. We should add an optional BY NAME clause, such that:

INSERT INTO tgt BY NAME <query>

takes each column of <query> and inserts it into the column in `tgt` which has the same name according to the configured `resolver` logic.

Issue Links

• links to: [Github] Pull Request #41389 (jose-torres)
• links to: [Github] Pull Request #41496 (Hisoka-X)
• links to: GitHub Pull Request #41389