drexler-sky opened a new pull request, #18885: URL: https://github.com/apache/iceberg/pull/18885
Related to #18235. `SchemaUpdate` renames or removes name-keyed per-column table properties when a column is renamed or dropped. Two cases were missed: 1. `write.parquet.bloom-filter-fpp.column.`, `write.parquet.bloom-filter-ndv.column.`, and `write.parquet.dict-encoding-enabled.column.` were not registered, so they were never updated. After a drop and re-add, the old setting applied to the new column. 2. Only the field that was directly renamed or dropped was handled. Properties on nested fields of a renamed or dropped struct (e.g. `location.lat`) were left behind. A stale `write.metadata.metrics.column.` property also made later property updates fail validation. This PR registers the missing prefixes and compares all field names between the old and new schema, so nested fields are included. This does not change the keying to field IDs, which #18235 proposes. -- 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]
