Phillip Cloud
Phillip Cloud
I'll fix any xfailing tests here and then merge.
Ok, fixed up the xfails and implemented the count distinct star with filter case.
The tricky bit is that `COUNT(DISTINCT ...)` can't use the usual aggregation filter syntax, so you have to do the "null if the filter is true" thing.
BigQuery is passing: ``` …/ibis on ssabdb/main:main is 📦 v9.1.0 via 🐍 v3.10.14 via ❄️ impure (ibis-3.10.14-env) ❯ pytest -m bigquery -n 8 --dist loadgroup -q bringing up nodes......
Thanks @ssabdb, keep 'em coming!
Part of the reason that pandas API is so large and complex to use and maintain (in my opinion) is that it favors flexibility over consistency. I think we should...
> To confirm, I am assuming that `array_of_arrays = ibis.array([[ibis.literal(1)], [ibis.literal(2)]])` is a valid input. You also want to be able to support this, right? No, I'm saying we shouldn't...
@NickCrews I am still having a lot of trouble following everything here. I suspect that the PR is still trying to do too much. Can you provide a minimum list...
Looks like a bug, I'll poke around a bit to confirm!
I will clean up the `ruff` lints in #9489!