pvary commented on PR #10695: URL: https://github.com/apache/iceberg/pull/10695#issuecomment-2227695699
@fengjiajie: This change was introduced, because in older Iceberg versions the different statistics behaved differently. See: https://github.com/apache/iceberg/pull/7643 The goal is that the `DataFile` should be immutable. I have 2 main issues with the proposed change: - It make the statistics mutable - It would introduce an extra copy of the statistics maps, which is problematic from performance reasons. I think the main issue is that in your code the generated statistics maps are not stored in `SerializableMap` and `SerializableByteBufferMap` objects. This is the reason why the copy method falls back wrapping it to `Collections.unmodifiableMap(map)` instead of calling `map.immutableMap()`. The result of the former is not serializable with Kryo, while the result of the later should work. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
