alexeykudinkin commented on issue #5519: URL: https://github.com/apache/hudi/issues/5519#issuecomment-1307831723
Folks, this is unfortunately an issue not of Hudi itself, but rather w/in both Parquet and Spark: as part of holistically addressing issues of handling schemas in Hudi in https://github.com/apache/hudi/pull/6358, i've stumbled upon this issue as well and here are my findings: - Type promotions could only work properly (in Spark/Parquet env) in case you're overwriting all of your data-files. In case you have a heterogeneous env where you have both base files in old schema and the new schema you'll be having issues reading such table (see below) - Heterogeneous table w/ type promotions won't work w/ Parquet when Dictionary Encoding is enabled (requires fixing DE impl in Parquet) - Heterogeneous table w/ type promotions won't work w/ Spark Vectorized Reader either ([SPARK-40876](https://issues.apache.org/jira/projects/SPARK/issues/SPARK-40876?filter=allopenissues)) As such, there's no way this could be reliably supported w/in Hudi currently. Closing this for now. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
