Sum of null values

[zueblin]

When calculating the sum of a column containing only null values, the result is 0. To me this seems wrong.
To reproduce:

val age_int = listOf(null, null).toColumnOf<Int?>("AGE")
println(dataFrameOf(age_int).sum("AGE"))

Output:

0

The documentation states that the sum operation ignores null values. I would expect the result to be null instead of 0.
Is this a bug, or does it work as designed? If its not a bug, would it be possible to control the behaviour in such a case? Or am i missing possible workarounds?

Thanks for you ideas!

[Jolanrensen]

It works as intended. The sum operation "ignoring" null values means it interprets them as the zero-value of the type you give it. So since you create a column of Int?, the zero value becomes 0.

If we don't do this by default it's ambiguous what "summing nulls" means. This is also why in the standard library, there are no .sum() functions for nullable iterables, just .sumOf {} which expects you to replace nulls.

So, if you expect "all nulls" to become null, you could write something like:
val sumOrNull: Int? = if (age_int.allNulls()) null else age_int.sum()

or as extension function on a DataColumn:

fun <C : Number> DataColumn<C?>.sumOrNull(): C? =
    if (allNulls()) null else sum()

[zueblin]

Many thanks for the explanation and suggested workarounds. I gave the extension function a quick try and i think i will be able to solve my problem with this approach.