JeonDaehong opened a new issue, #3367: URL: https://github.com/apache/iceberg-rust/issues/3367
### Is your feature request related to a problem or challenge? #3329 adds a sorting position-only delete writer. Like the current writer, it puts the deletes for all data files it sees into one delete file (PARTITION-style). To get one delete file per data file (FILE-style), the caller has to create a writer per data file. In Java, `SortingPositionOnlyDeleteWriter` takes a `DeleteGranularity` (`FILE` / `PARTITION`) and does that split itself. The table property is `write.delete.granularity` (default `partition`), and Spark writes default to `file`. File-scoped delete files can be matched by path (#2936), so they are cheaper to apply on read. This came up in #3329 (from the MOR native write work in datafusion-comet). ### Describe the solution you'd like - Add a `DeleteGranularity` enum (`File`, `Partition`) and read `write.delete.granularity` from table properties. - Let the sorting position-only delete writer from #3329 take it and, for `File`, write one delete file per referenced data file, as Java does. This depends on #3329. ### Willingness to contribute I can contribute to this feature independently -- 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]
