jackylee-ch opened a new pull request, #12737: URL: https://github.com/apache/gluten/pull/12737
Test-only. Makes gluten use the Arrow the Spark distribution already provides instead of bundling and partially shading its own copy. - `gluten-arrow`: `arrow-memory-*` / `arrow-memory-core` / `arrow-vector` move to `provided`; `arrow-c-data` / `arrow-dataset` exclude them from their transitive closure. - `backends-velox`: re-declares the three `provided` deps (`provided` is not transitive). - `package`: drops the `org.apache.arrow` relocation block. `arrow-c-data` / `arrow-dataset` stay bundled and unrelocated — their native libraries bind JNI symbols to the original package names. `arrow.version` is unchanged (3.3/3.4/3.5 = 15.0.0, 4.0/4.1 = 18.1.0). Bundle: 138.6 MB → 130.9 MB compressed; `arrow-vector` 723 → 3 classes, `arrow-memory` / `flatbuf` / `flatbuffers` → 0. Spark 3.3 lanes are expected to fail: bundled `arrow-c-data` needs `org.apache.arrow.memory.ForeignAllocation` and `org.apache.arrow.vector.table.Table`, both introduced in Arrow 11, while Spark 3.3 ships Arrow 7.0.0. Spark 3.4 / 3.5 / 4.0 / 4.1 resolve with no missing classes. ### Was this patch authored or co-authored using generative AI tooling? Generated-by: Claude claude-opus-4.7 -- 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]
