Weixin-Xu opened a new pull request, #13492: URL: https://github.com/apache/gluten/pull/13492
## What changes are proposed in this pull request? Related to #12998. Restore the Bolt backend's end-to-end build and execution path so users can build native libraries, package the Gluten JAR, and run Spark queries with Bolt. - Adapt Arrow batch handling to the current vector APIs and preserve independent ownership for each Python output batch. - Align Uniffle writer and JNI interfaces, including accounting for raw partition sizes after the final flush. - Fix native linkage and update Delta scan and spill-path API usage. - Update test sources required by normal JAR packaging and retain focused regression coverage for the runtime fixes. Dedicated UT integration, blacklist configuration, and CI enablement are deferred to a follow-up PR. ## How was this patch tested? - Built candidate native libraries with native tests, examples, and benchmarks disabled. - Successfully ran `make jar_spark35` and seven additional Maven packaging configurations covering Spark 3.4–4.2, Scala 2.12/2.13, and selected JDK and integration profiles. - Verified that all eight packaged JARs contain the candidate native libraries and the expected Bolt revision (`c3548fe`). - Started Spark 3.5.5 with the generated JAR and ran queries against real Parquet files covering scan/filter/projection, aggregation with shuffle, join with aggregation, and ordering with a limit. Results matched vanilla Spark, and execution plans were checked for native operators. - Passed two Arrow output-batch ownership regression tests. Runtime validation was performed on Spark 3.5.5; the other configurations were validated through packaging. The full backend UT matrix was not run. ## Was this patch authored or co-authored using generative AI tooling? Generated-by: OpenAI Codex (GPT-6) -- 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]
