gszadovszky commented on PR #184: URL: https://github.com/apache/parquet-format/pull/184#issuecomment-1592472334
I think that compatibility in Parquet file format is such a strong requirement that extending primitive types is simply not an option. (I agree though, if I would introduce a new file format I would specify only 3 primitive types: FIXED_LEN, VAR_LEN, BIT. But we already have what we have.) Meanwhile, I don't see why the Float16 physical type would be a requirement to use BYTE_STREAM_SPLIT. I don't think it is widely used so we can update the related spec to allow this encoding to be used for any primitive type (except BOOLEAN). Then, it is up to the implementations to use it for `FIXED_LEN_BYTE_ARRAY[2] (FLOAT16)`. -- 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]
