[ https://issues.apache.org/jira/browse/ARROW-91?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15275294#comment-15275294 ]
Uwe L. Korn commented on ARROW-91: ---------------------------------- [~wesmckinn] I had a look at your FileReader API draft and I'm not sure if that is the one I would implement as a first draft. As a minimal first interface, I would implement a function/class that takes a memory pool and a `ParquetFileReader` and then simply returns the Arrow structure. Although this hides a lot of the flexibility, this is probably the most common use case. From there on, we could add more flexibility to the arrow<->parquet library. > C++: First draft of an adapter class for parquet-cpp's ParquetFileReader that > produces Arrow table/row batch objects > --------------------------------------------------------------------------------------------------------------------- > > Key: ARROW-91 > URL: https://issues.apache.org/jira/browse/ARROW-91 > Project: Apache Arrow > Issue Type: Improvement > Components: C++ > Reporter: Wes McKinney > > This will be the first step in exposing Parquet file data to Arrow consumers > with a nice C++ API. -- This message was sent by Atlassian JIRA (v6.3.4#6332)