Thank you @HuangZhenQiu.

I created a [(quick 
patch),](https://github.com/lvhuyen/flink/tree/parquet_input_format(7243)) for 
that Timestamp conversion and my Flink job has been running well with 
ParquetPojoInputFormat.

One more issue I found in the current build is with the parquet file having a 
column of type Array of primitive. As per parquet format spec, schema for this 
column will have 2 layers of GroupType, while in the method 
_ParquetSchemaConverter.convertField(final Type fieldType)_, only one is 
handled. So, in the result, we'll not get an array of primitive types, but one 
ObjectArray which has only 1 item.
This issue is with ParquetMapInputFormat only, so I have not done anything with 
it in my branch.

[ Full content available at: https://github.com/apache/flink/pull/6483 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to