Support QBit data type (25.10)

[mshustov]

Website docs: https://clickhouse.com/docs/sql-reference/data-types/qbit

Internal layout: https://github.com/ClickHouse/ClickHouse/pull/87922/files#diff-9d45fe76d9e7f9a392db684d3465aff14979cfc9bbe5ccd912e9307a934bd28aR15-R51

Given QBit(element_type, dimension) where element_type is one of BFloat16, Float32, Float64, it is internally represented as a Tuple of FixedString(dimension).

• The length of the Tuple depends on the element_type
  • BFloat16 = 16
  • Float32 = 32
  • Float64 = 64
• Requires allow_experimental_qbit_type setting

[Shofiya2003]

Is someone working on this issue? How can I get started with this? @mshustov

[joe-clickhouse]

Hi @Shofiya2003, you're more than welcome to tackle it! I'd start by familiarizing yourself with the QBit type in the links provided above. As for implementation in the codebase, you'll want to take a look in https://github.com/ClickHouse/clickhouse-connect/tree/main/clickhouse_connect/datatypes as that's where all the datatype classes are implemented. Maybe put it in its own file called vector.py. Decide on what native python type it should map to (maybe a list of floats)? The key methods you'll need to implement will be _read_column_binary and _write_column_binary. The BFloat16 might be good to reference. I'm not super familiar with the type yet myself so you'll need to explore a little bit with how it's sent over the wire etc. Please feel free to reach out with questions or issues if you run into any!