#890 - Cannot call .map {} on ColumnGroup

[ysknsid25]

Fix issue #890

If you want to use DataFrame<T>.map(), you can usually just call it with the default arguments. If you have something else you want to transform, I don't mind returning the result of passing it to you.

before

after

[Jolanrensen]

Thank you very much!

It does, however, seem that your fix breaks the notation:

df.map { ... }

as seen in the test here: https://github.com/ysknsid25/dataframe/blob/b67c665cb10375ac13f2dd9bca8aa838af522082/core/src/test/kotlin/org/jetbrains/kotlinx/dataframe/samples/api/Modify.kt#L1130

I think it may be easier to just provide another map function but defined on ColumnGroup instead of DataFrame. I suspect this will work, because the compiler always chooses the function that matches the provided type the closest. Could you try?

[ysknsid25]

I defined the following in map.kt, but it did not resolve the ambiguous overload 🫠

// region ColumnGroup

public fun <T, R> ColumnGroup<T>.map(transform: (AnyCol) -> R): List<R> = columns().map { transform(it) }

// endregion

[Jolanrensen]

I see :( that's a shame.

Then the only solution may be to add

public fun <R> map(transform: RowExpression<T, R>): List<R> = asDataFrame().map(transform)

directly inside the ColumnGroup interface. Member functions do always have priority over extension functions, so it would probably work fine. (just need to check if it doesn't break anything else :) )

[Jolanrensen]

I see :( that's a shame.

Then the only solution may be to add

public fun <R> map(transform: RowExpression<T, R>): List<R> = asDataFrame().map(transform)

directly inside the ColumnGroup interface. Member functions do always have priority over extension functions, so it would probably work fine. (just need to check if it doesn't break anything else :) )

Another would be to use @HidesMember from https://github.com/JetBrains/kotlin/blob/master/libraries/stdlib/src/kotlin/internal/Annotations.kt