gunnarmorling opened a new pull request, #128: URL: https://github.com/apache/parquet-testing/pull/128
Hey, this PR adds two fixtures for the new `FILE` logical type ([apache/parquet-format#585](https://github.com/apache/parquet-format/pull/585), [#603](https://github.com/apache/parquet-format/pull/603)). - _data/file_reference.parquet_ (14 rows): one `FILE` group with all six children, one reference form per row. It covers inline bytes, empty inline bytes, a whole external object, a prefix (`uri` + `size`), an offset range, a zero-length range, inline bytes together with a locator, a relative URI, one checksum per algorithm (`MD5`, `CRC32`, `CRC32C`, `SHA-256` over `hello`, so each digest can be checked), an `ETAG`, and a null group. - _data/file_reference_shapes.parquet_ (3 rows): the shapes the spec allows besides the full group. It covers the spec's two example subsets, children in non-spec order, a `REQUIRED` group, and `FILE` groups as list elements and map values, including empty and null containers. Both are uncompressed and were written by [Hardwood](hardwood.dev) (commit https://github.com/hardwood-hq/hardwood/commit/ad2eef066ae668a72c0693a827576b4cc675ea36, it's going to be part of 1.1.0.CR1), where we added support for `FILE` in PR see [hardwood-hq/hardwood#1494](https://github.com/hardwood-hq/hardwood/pull/1494); the sketch for the generator is in `file_reference.md`. parquet-java 1.18.1, which predates `FILE`, reads every group back as a plain group with the documented values. It's the first time I'm opening a PR here, so any guidance on required shape/form/protocol would be appreciated very much. Thanks a lot! _(written with the help of Claude, but heavily editorialized)_ -- 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]
