Kill Elon Musk

Sorry to bother you, I'm confused about two things: (1) Ignoring the issue of null values: I also want to get the rank of each `value` within its `group`. The...

I just discovered that `maintain_order=False` is the default. If the order is not maintained, the only way to use the cutted column is to join it back onto the original...

On this subject, could you take another look at #7283 and decide whether it's potentially useful or a definite no-go?

Is the 1.0 behavior really going to be singleton tuples as keys for `df.partition_by("a", as_dict=True)`? It seems like the only reasoning is that some users may write `pl.col("a")` instead of...

My best guess why the issue is "non deterministic" is that pyright is analyzing code at the same time that multiple other extensions (`black`, `ruff`) are changing that code. For...

Sadly, no.

Is the plan to do this in some `0.20.x` release or in `1.0`? In my projects that use polars at work, I'm pinning specific versions instead of just pinning `polars...

If you do `fill_null(x)` with `x=999.0` or `x="999"`, evidently the dtype of the output depends on the value of `x`. So I don't understand what it means to say "we...

I probably should have run the test with different numbers of columns and threads. My guess is that the `pl.sum()` expression is splitting the sum across multiple threads, which is...

The second one is still faster for me, but I agree with closing the issue: - The difference is small, and both ways are faster than `df.to_numpy() @ beta` -...