具有2个泛型的scala方法，可以推断出一个

Question

I'm new to Scala. I am writing a method which sends a request of type RQ to a server, and receives a response of type Response[RSP]. The simplified version: I define a method to make a call:

def invoke[RQ, RSP](request: RQ): Response[RSP] = {....}

When I invoke the method, all parameters can be inferred if I specify the type of the receiving variable, like this:

val result: Response[Double] = invoke("Hello")

But if I just say invoke("Hello") without assigning it to anything, it won't compile. I have to supply the 2 generic types (eg invoke[String, Double]("Hello") ). My question is: The String can be inferred from the parameter. Is there a way to write the call specifying only the RSP generic? Something in the lines of : invoke[ , Double]("Hello") ?

Answer 1

No.

This is a feature that has been asked about a number of times. Future versions of Scala may support a syntax similar to named arguments for methods:

invoke[RSP = Double]("Hello")

At the moment, there is nothing much you can do except restructure your code so that the method has only one type parameter, and the second comes from the surrounding context.

The more interesting question is: why do you even care about the return type if you just throw away the result anyway? It looks like your method violates the Command-Query Separation Principle. If your method simply returned Unit , you wouldn't need the second type parameter, and all would work fine. If you actually used the return value, the type inferencer would at least have a chance to infer the second type parameter from the context. But the fact that you have a type parameter for the return value but ignore the return value means that there is no context for the type inferencer to look at.