I am attempting to troubleshoot and provide a patch if I am capable for ARROW-2367 (https://issues.apache.org/jira/browse/ARROW-2367). From what I can tell from gdb on a debug build of master, I believe the issue to be lists in individual rows in an Pandas dataframe/series being stored as a single BinaryArray instead of a ChunkedArray when the size of the total column data exceeds the max int32 size.
How would I confirm this hunch? Apologies if this something straightforward; new to the project and this is my first time debugging a Python C/C++ extension. Thanks, Bryant