How to interpret this scala syntax "Class[_ >: Float with Int with Double]"

Question

When I read Mxnet source code, I was confused in following statements:

object NDArray {
  private val logger = LoggerFactory.getLogger(classOf[NDArray])
  private[mxnet] val DTYPE_NATIVE_TO_MX: Map[Class[_ >: Float with Int with Double], Int] = Map(
    classOf[Float] -> 0,
    classOf[Double] -> 1,
    classOf[Int] -> 4
  )

What does it mean for "Class[_ >: Float with Int with Double], Int]"? I understand the scala keyword "with" could be used during class declaration, for example

Class person with glass {

means the class 'person' has the trait of objdect 'glass'.

How to interpret the usage of 'with' in above code?

score 2 · Answer 1 · answered Oct 12 '16 at 11:29

2

The with keyword is used for expressing intersection types.

The type Float with Int with Double is basically a subtype of Float and Int and Double. Of course you cannot have an actual value of this type because Float, Int and Double are all final classes. Here, in the type Map[Class[_ >: Float with Int with Double], Int], it is used to express that every key of the Map has to be a Class[T] where T has to be a supertype of Float with Int with Double. And those supertypes are Float, Int and Double (and AnyVal and Any, if we go higher up the inheritance chain).

answered Oct 12 '16 at 11:29

Jasper-M

14,966
2
26
37

So could it also be a string, because String is a subtype of Any? – Madoc Oct 12 '16 at 12:08
What could be a string? There are only `Class` objects here. – Jasper-M Oct 12 '16 at 12:21
Sorry, I meant the String class. – Madoc Oct 12 '16 at 23:36
2

`Class` is invariant so normally not. `val c: Class[_ >: Float with Int with Double] = classOf[String]` doesn't compile. – Jasper-M Oct 13 '16 at 06:28

How to interpret this scala syntax "Class[_ >: Float with Int with Double]"

1 Answers1