jackylee-ch opened a new pull request, #853: URL: https://github.com/apache/paimon-rust/pull/853
`parse_compression` fell back to `Compression::UNCOMPRESSED` for any unrecognized `file.compression`. Nothing on the plain-parquet path re-checks the codec, so `gzip2` — or a stray space in `'zstd '` — wrote raw pages and recorded codec id 0 in the footer, leaving the finished file indistinguishable from `file.compression=none`. Java rejects the same value where the per-file writer is built: `RowDataParquetBuilder#createWriter` -> `CompressionCodecName#fromConf`. Accepted: `none, uncompressed, snappy, gzip, gz, lz4, lz4_raw, zstd`. `lz4_raw` is added because Java accepts it, so a Java-created table can carry it — that also changes its output from uncompressed to LZ4_RAW. `brotli` and `lzo` are rejected: no Java codec class ships for either, and `LZO` has no arrow-rs encoder. A Rust-only table with a typo'd codec now fails on the first write; Java would already have rejected it. -- 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]
