coastalwhite commented on code in PR #153: URL: https://github.com/apache/parquet-site/pull/153#discussion_r2790279767
########## data/implementations/support/polars.yaml: ########## @@ -0,0 +1,127 @@ +engine_id: polars +last_updated: "2026-01-30" +support: + physical-boolean: + status: full + physical-int32: + status: full + physical-int64: + status: full + physical-int96: + status: read + physical-float: + status: full + physical-double: + status: full + physical-byte-array: + status: full + physical-fixed-len-byte-array: + status: read + logical-string: + status: full + logical-enum: + status: read + logical-uuid: + status: read + logical-int-types: + status: full + logical-decimal-int32: + status: read + logical-decimal-int64: + status: read + logical-decimal-byte-array: + status: read + logical-decimal-fixed-len-byte-array: + status: full + logical-float16: + status: full + logical-date: + status: full + logical-time-int32: + status: read + logical-time-int64: + status: full + logical-timestamp-int64: + status: full + logical-interval: + status: read + logical-json: Review Comment: Looking into it for a second. [Polars indeed should fall through to BinaryView transparently.](https://github.com/pola-rs/polars/blob/3a37c4145ef4f4d46746b08cb5e57757f246ecd2/crates/polars-parquet/src/arrow/read/schema/convert.rs#L164) I have never myself encountered a JSON or BSON file in the wild, so I have never tested it, nor do I remember ever having dealt with any issues about it. (PyArrow, which was my general way of generating Parquet files with specific parameters, also does not support generating them, AFAIK.) My general rule of thumb is: if it is not structurally tested, it probably doesn't work. I think, therefore, there is a high chance Polars does not handle these correctly at the moment. But I can adjust the matrix to show that the *intent* is that it should work. So that people will file a bug report instead of a feature request. I am fine merging like this as well, though. -- 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]
