ccat3z opened a new pull request, #8861: URL: https://github.com/apache/incubator-gluten/pull/8861
## What changes were proposed in this pull request? If `orc.force.positional.evolution` is enabled (https://github.com/apache/spark/commit/00d43b1f829fb5f79f0355afbbacc804162648e5) or the orc file has no column names, vanilla spark read orc file using table schema by index. 1. Passthrough `ScanTransformer#getDataColumns` as table schema to Velox 2. Enable `k{Parquet,Orc}UseColumnNames` and `kOrcUseNestedColumnNames` in Velox to match spark default behavior, which always map table schema to physical file schema using name. 3. Set `force_positional` if `orc.force.positional.evolution` is enabled. This PR depends on https://github.com/facebookincubator/velox/pull/12489 (old orc file) and https://github.com/facebookincubator/velox/pull/12490 (readOptions `force_positional`). Fixed https://github.com/apache/incubator-gluten/issues/5638. ## How was this patch tested? UT `VeloxOrcPositionalColumnsSuite`. -- 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]
