You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Mark Bathori (Jira)" <ji...@apache.org> on 2025/02/12 12:31:00 UTC

[jira] [Created] (NIFI-14260) ParquetReader can't handle timestamps stored as INT96 type

Mark Bathori created NIFI-14260:
-----------------------------------

             Summary: ParquetReader can't handle timestamps stored as INT96 type
                 Key: NIFI-14260
                 URL: https://issues.apache.org/jira/browse/NIFI-14260
             Project: Apache NiFi
          Issue Type: Improvement
            Reporter: Mark Bathori
            Assignee: Mark Bathori


There are components e.g.: Spark that still stores timestamp values in INT96 type. Currently the ParquetReader can read these type of values only as Byte Array so they can't be handled as timestamp. Since the INT96 type is not exclusively used for timestamp fields and the schema does not contain any information about the original type of these fields so a similar approach could be added as int the ParquetWriter.

The idea is to add a new property where the user can list all the INT96 fields that should be handled and parsed as timestamp.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)