[PARQUET-2163] Parquet C++ Float Runtime Error in Decimal Schema

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: cpp-9.0.0
• Component/s: parquet-cpp
• Labels: pull-request-available
• Assignee: William Butler
• Reporter: William Butler
• Created: 28/Jun/22 22:09
• Updated: 23/Jun/24 03:32
• Resolved: 06/Jul/22 16:02

Description

The code that determines if a physical type is appropriate for a particular decimal logical type can overflow to infinity which then triggers a floating point error when converting to int32.

 

I don't know how large of decimals we want to support but there is a simple fix to support up to ~2^31 -1 digits.

Issue Links

• links to: GitHub Pull Request #13456