voonhous commented on PR #19144: URL: https://github.com/apache/hudi/pull/19144#issuecomment-4873112656
CI red on the last push was a real regression, fixed in the latest commit: Avro `Schema.equals` ignores doc strings and aliases, so value-interning inside `fromAvroSchema` collapsed a commented table schema onto a previously-interned uncommented-but-equal wrapper -- hive sync then built column comments from empty `doc()`s (`testSyncWithCommentedSchema` / `testUpdateTableComments`). Fix: the canonical pool now keys on the schema's full serialized content (plus wrapper class), which does cover docs/aliases -- interning is never lossy, while content-identical schemas still dedup. Added a unit test pinning doc/alias preservation through `fromAvroSchema`; the six failing hive-sync tests pass locally against the fix. `HoodieSchema.equals` still follows Avro semantics (doc/alias-blind) -- now documented on the method. -- 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]
