kamcheungting-db commented on PR #1062: URL: https://github.com/apache/iceberg-cpp/pull/1062#issuecomment-6791426731
> IMHO, I don't think this is the right direction. It makes unnecessary copies of schemas even when we don't have any list columns. I think a better approach would be asking file format readers to always respect `read.arrow.use-large-list` explicitly. For now, we can simply error out if the expectation is not met. WDYT? Ans: Agreed. I reject the invalid schema. In the new implementation, the `read.arrow.use-large-list `is enforced by the file format readers themselves, on every projected list (including nested ones). The Parquet reader validates the field, when it conflicts with the configured width, list widths are never converted. The Avro reader errors out when use-large-list=true meets any projected list. Test cases are also added for this new behavior. -- 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]
