Polars support

[axelwas]

Hello, thank you for this great library! Is it planned to support Polars library at some point ?

[genzgd]

Polars is not currently on the roadmap, but we do look at the community interest in new features on a regular basis and adjust our priorities accordingly. Of course we'd welcome a pull request from someone who is familiar with Polars.

[qqletter]

Yes, I do hope this client will support polars.

[NotSimone]

You can query to get pyarrow tables, which can then zero-copy into Polars with something like this:

df = pl.from_arrow(client.query_arrow(query))

[genzgd]

Feedback based on @NotSimone's comment would be appreciated -- is using PyArrow enough of a solution or does Polars use some other internal data format?

[axelwas]

Polars does use Pyarrow under the hood so I think this would work.

Since my question, we ended up sticking with Pandas. Would it be possible to optionally make query_df return a Pyarrow backed Pandas DataFrame instead of Numpy backed ? I see use_numpy param is hardcoded to true here.

[genzgd]

Agreed that PyArrow DataFrames would be useful - https://github.com/ClickHouse/clickhouse-connect/issues/326. However I'm not sure when it will it will be prioritized.