c-thiel opened a new pull request, #2880: URL: https://github.com/apache/iceberg-rust/pull/2880
## What changes are included in this PR? Project variant columns on read by mapping their metadata/value leaves (which carry no field id — only the enclosing group does) to the variant's field id, mirroring Java's PruneColumns: the whole group is projected with no type-promotion check. Shredded variants (typed_value present) are rejected until we can reconstruct them. collect_variant_leaves numbers leaves exactly like Fields::filter_leaves, including its Dictionary/RunEndEncoded unwrapping and Union descent, so a sibling of those kinds can't shift a variant's leaves onto the wrong column. The pruned schema used for the type-promotion check includes variant leaves too, keeping map<_, variant> a well-formed two-field map. Variant identity is recovered from the Iceberg schema: fields the table schema declares as variants are tagged with the arrow.parquet.variant extension so arrow_schema_to_schema folds the storage struct back into Type::Variant. This mirrors iceberg-java's TypeWithSchemaVisitor, which recognizes a variant by the Parquet annotation OR the Iceberg type, and avoids depending on the Parquet variant annotation being present. ## Are these changes tested? yes ----- Based on https://github.com/apache/iceberg-rust/pull/2188 and https://github.com/apache/iceberg-rust/pull/2840 and -- 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]
