refactor!: rename <Expr>$lit_to_s() to <Expr>$to_series() and remove <Expr>$lit_to_df()

[eitsupi]

• These functions do not exist in py-polars
• Their names differ from the rules for other functions (should be $to_* instead of $lit_to_*)
• It is ambiguous whether "df" represents data.frame or RPolarsDataFrame

So I think it is better to remove them.

[etiennebacher]

This is fine for me but I'd like @sorhawell to take a look too

[sorhawell]

I disagree with dropping lit_to_s().

How about keeping lit_to_s() or renaming to to_s()?

I find it very practical and is one of my favorite methods. All the tests become more verbose by using as_polars_series() and it is less easy to fast tab-autocomplete and it breaks the polars syntax if required to use |> for conversion.

These functions do not exist in py-polars

Neither do as_polars_series() right?

Their names differ from the rules for other functions (should be $to_* instead of $lit_to_*)

That is because only the subvarient Expr::Literal or (Expr derived thereof) can be converted into a Series. I prefixed lit_ point this out.

#example of non literal conversion to Series
> as_polars_series(pl$col())

Error: Execution halted with the following contexts
0: In R: in $select()
0: During function call [as_polars_series(pl$col())]
1: Encountered the following error in Rust-Polars:
not found:

Error originated just after this operation:
DF []; PROJECT */0 COLUMNS; SELECTION: "None"

We can drop the prefix lit, but documentation should state that it will cause an error converting non Literal-derived expressions to Series. We can consider change to to_s().

It is ambiguous whether "df" represents data.frame or RPolarsDataFrame

agreed but that could be resolved by renaming to lit_to_xyz. I don't think lit_to_df is as important as lit_to_s().

[sorhawell]

a use case for lit_to_s(): If working with a Series, we have no cast. to_lit() and lit_to_s() allows jumping back and forth with method chaining only.

s = pl$Series[1:3]
s$
  to_lit[]$
  # work with a Series as a Literal
  cast[pl$Float64]$
  lit_to_s[]$
  # back to Series interface
  append[4]$
  chunk_lengths[]
[1] 3 1

sry for hard brackets I got my self hooked tab-complete heaven :)

[eitsupi]

I see.
Then how about deleting lit_to_df and renaming lit_to_s to to_series?

[sorhawell]

I see. Then how about deleting lit_to_df and renaming lit_to_s to to_series?

sounds good to my ears :)

[eitsupi]

@sorhawell I have updated. Could you take a look at this?

[sorhawell]

looks good to me :)