Dale Jung comments

Results 17 comments of


                                            Dale Jung

CSS Grid Compatibility

So I think I have css grids working for the most part. I had to turn off height clamping because the grid items would expand if we filtered out enough...

.load breaks on multiline files

Oh, right. `.load` comes from base node repl.

`getattr` in child class gets called when mixed with `cached_property`

I just checked against latest master and it still exists. It looks like 23.2.0 is when the error started, 23.1.0 works.

pl.Expr.get(pl.Expr.arg_min()) does not respect rolling window

This also appears to be an issue for `group_by_dynamic` when defining different `every` and `period`. ```python dates = pl.date_range( pl.datetime(2023, 1, 1), pl.datetime(2024, 1, 1), eager=True, ) df = pl.DataFrame({...

pl.Expr.get(pl.Expr.arg_min()) does not respect rolling window

Possible that internally it is the same bug as: https://github.com/pola-rs/polars/issues/15091 I thought `group_by_dyamic` behaved better but now seeing that `offset` will return incorrect results like `rolling`, it feels similar.

pl.Expr.get(pl.Expr.arg_min()) does not respect rolling window

This looks to be fixed now.

Allow `pl.Expr.get` handle null index values in Aggregation context so that `pl.Expr.arg_min/arg_max` is always valid input

@orlp sounds good. Yeah I originally opened this ticket thinking my particular issue was with OOB but realized it was the nulls.

Allow `pl.Expr.get` handle null index values in Aggregation context so that `pl.Expr.arg_min/arg_max` is always valid input

@orlp I still get out of bounds error for ```python report = df.group_by_dynamic('date', every='1mo').agg( pl.col('num').filter(after_28).arg_min(), # THIS ERRORS WITH OutOfBoundsError pl.col('date').filter(after_28).get( pl.col('num').filter(after_28).arg_min() ).alias('after_28_date'), ) ``` I just realized that this...

Allow `pl.Expr.get` handle null index values in Aggregation context so that `pl.Expr.arg_min/arg_max` is always valid input

@orlp Looking at `take.rs` I see that we'd run into the same issue with `slice`. It's interesting, a `slice(0, pl.col('col').arg_max())` would fail but `slice(0, 0)` would work fine. Furthermore a...

Allow `pl.Expr.get` handle null index values in Aggregation context so that `pl.Expr.arg_min/arg_max` is always valid input

@ritchie46 @orlp Can I get clarification on whether this is a bug or usage issue? Is the fix adding `null_on_oob` to `Expr.get`? Or should nulls not trigger `OutOfBoundsError`.