Split move {}.toTop {} into toTop {} and toTop() for correct compiler plugin support #1508
Description
We have
@Refine
@Interpretable("ToTop")
public fun <T, C> MoveClause<T, C>.toTop(
newColumnName: ColumnsSelectionDsl<T>.(ColumnWithPath<C>) -> String = { it.name() },
): DataFrame<T> = into { newColumnName(it).toColumnAccessor() }However, no matter how it's called, the compiler plugins calls toTop { it.name() }.
Meaning, if a user specifies newColumnName like move { group.col }.toTop { it.name() + "1" } .col is available on the result, but at will fail at runtime as there's only the "col1" column.
We cannot reliably see what's inside this lambda expression (unless we follow rename { singleCol }.into { "string constant" }), so it may be best to split it:
@Refine
@Interpretable("ToTop")
public fun <T, C> MoveClause<T, C>.toTop(): DataFrame<T> = toTop { it.name() }
public fun <T, C> MoveClause<T, C>.toTop(
newColumnName: ColumnsSelectionDsl<T>.(ColumnWithPath<C>) -> String,
): DataFrame<T> = into { newColumnName(it).toColumnAccessor() }This doesn't even need changes in the compiler plugin AFAIK :)