[ARROW-17524] The ORC reader method ReadStripe does not work when we specify fields to selected as a list of integers

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 8.0.1
• Fix Version/s: 10.0.0
• Component/s: C++
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/pull/13962

Language

C++

• Assignee: Louis Calot
• Reporter: Louis Calot
• Created: 25/Aug/22 13:43
• Updated: 11/Jan/23 11:51
• Resolved: 18/Oct/22 20:27

Description

I think there is a bug in the ORC reader : when we specify the fields indexes that we want to keep, it does not work correctly. Looking at the code, it seems to be because we do "includeTypes" in lieue of "include" when setting the ORC options.
It can be problematic when we want to import an ORC table containing Union types as it will do an error at the import, even if we try not to import these specific fields.

The definitions of the corresponding ORC methods are here :
https://github.com/apache/orc/blob/72220851cbde164a22706f8d47741fd1ad3db190/c%2B%2B/src/Options.hh#L185-L191

and
https://github.com/apache/orc/blob/72220851cbde164a22706f8d47741fd1ad3db190/c%2B%2B/src/Options.hh#L201-L207

Issue Links

• links to: GitHub Pull Request #13962