cxzl25 opened a new pull request, #56591: URL: https://github.com/apache/spark/pull/56591
### What changes were proposed in this pull request? This PR mirrors what `buildColumnarReader` already does since SPARK-44556: capture the `readerOptions` returned by `createORCReader`, then pass `readerOptions.getOrcTail` when constructing the per-split record reader so the footer is not re-read. ### Why are the changes needed? `OrcPartitionReaderFactory.buildReader` (the non-vectorized / row-based read path) previously called `OrcInputFormat.createRecordReader(fileSplit, taskAttemptContext)`, which internally calls `OrcFile.createReader` without an `OrcTail` and therefore re-parses the file footer from storage on every split. Without OrcTail reuse the non-vectorized path pays this cost a second time when opening the data reader for each split, while the vectorized path has been avoiding it since SPARK-44556. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? GHA ### Was this patch authored or co-authored using generative AI tooling? Generated-by: Claude Code -- 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]
