Phillip Cloud
Phillip Cloud
Thanks for the issue! I think table UDFs need their own issue and some discussion about the API, especially the bits around getting the schema of the return value of...
Closing as stale. Seems like there's a viable workaround for now.
**TL; DR:** queries *appear* faster because the polars versions are running against in-memory data, while the Ibis queries are running against `pl.scan_parquet`. The code in these notebooks is not close...
I just ran these all after converting `_scan_ds` to just return `scan` and the only query that was more than a factor of 2 worse in performance was query 2,...
Thanks for bringing this to our attention! It looks like there's something to look at for query 2!
@kunishou Awesome, great to hear! Let us know how we can help.
This looks to be caused by `pandas==2.2.0`, which breaks a bunch of timestamp related functionality. We have a bot PR (#8056) that I am slowly working through to try to...
It seems like there was a bunch of churn in supported datetime64 units: * 1.x: only `datetime64[ns]` supported * 2.1.x: additional units including `datetime64[D]` supported * 2.2.x: `datetime64[D]` no longer...
Fixed by #8758.
Hey all, I'm the lead developer of ibis! We're happy to support alternative methods of connecting to mssql-oriented DBs. We haven't added that capability yet because no one has asked...