LuciferYang commented on PR #3540: URL: https://github.com/apache/iceberg-rust/pull/3540#issuecomment-6910014319
Once #2906 lands, predicates on struct-nested leaves reach the row filter too; today they fail before it is built. With #2906 alone, `person.n < 3000000000` on a file that wrote `person.n` as `int`, read as `long`, returns no rows, like the top-level case here. I'd like your view on how the two fit together. `filter_schema` is built with `snapshot_schema.as_struct().field_by_id(id)`, which looks only at top-level fields, while the ids passed in are leaf ids. For `person.n` (id 3 under `person`) that returns `None`, so the filter schema has no `person` field. I think the transformer would then fail every nested-leaf predicate with `field not found` on the first batch, promoted or not, and #2906's nested tests would hit it after a rebase. Could you add a nested-leaf case to the promoted-column tests? -- 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]
