[GitHub] [spark] viirya edited a comment on pull request #33913: [SPARK-36669][BUILD] Revert to non-shaded Hadoop client library

viirya edited a comment on pull request #33913: URL: https://github.com/apache/spark/pull/33913#issuecomment-914990449 > Does this help for snappy-java? > > > Fixed the pure-java Snappy fallback logic when no native library for your platform is found. > > https://github.com/xe

[GitHub] [spark] viirya edited a comment on pull request #33913: [SPARK-36669][BUILD] Revert to non-shaded Hadoop client library

viirya edited a comment on pull request #33913: URL: https://github.com/apache/spark/pull/33913#issuecomment-913449357 > It works for parquet, but does it work for compressed lz4 hadoop seq file which will need full Lz4Codec support? > > Thanks, lz4-java APIs are only used int

[GitHub] [spark] viirya edited a comment on pull request #33913: [SPARK-36669][BUILD] Revert to non-shaded Hadoop client library

viirya edited a comment on pull request #33913: URL: https://github.com/apache/spark/pull/33913#issuecomment-913449357 > It works for parquet, but does it work for compressed lz4 hadoop seq file which will need full Lz4Codec support? > > Thanks, lz4-java APIs are only used int

[GitHub] [spark] viirya edited a comment on pull request #33913: [SPARK-36669][BUILD] Revert to non-shaded Hadoop client library

viirya edited a comment on pull request #33913: URL: https://github.com/apache/spark/pull/33913#issuecomment-913417449 Thanks @pan3793. I tried to add lz4 wrapper classes in #33912. Fortunately only a few lz4 APIs were used at Hadoop Lz4 codec internally. So the wrapper classes are