This is based on the assumption that existing readers check the two fields in this specific order (first the embedded metadata and falling back to the external one in case the former is missing). Which reader implementations fulfill this assumption? I suggest you to explicitly mention them. For example, if you have checked that parquet-mr and parquet-cpp follow this order, name them here.
[ Full content available at: https://github.com/apache/parquet-format/pull/110 ] This message was relayed via gitbox.apache.org for [email protected]
