[PARQUET-1392] [C++] Supply row group indices to parquet::arrow::FileReader::ReadTable

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: cpp-1.5.0
• Component/s: parquet-cpp
• Labels: pull-request-available
• Assignee: Uwe Korn
• Reporter: Uwe Korn
• Created: 19/Aug/18 13:03
• Updated: 23/Aug/18 13:18
• Resolved: 23/Aug/18 13:18

Description

By looking at the Parquet statistics, a user can already determine with its own logic which RowGroups are interesting for him. Currently we only provide functions to read the whole file or individual RowGroups. By supplying parquet::arrow with the RowGroups at once, it can better optimize its memory allocations as well as make better use of the underlying thread pool.

Issue Links

• relates to: PARQUET-1158 [C++] Basic RowGroup filtering
• links to: GitHub Pull Request #492