[转]Scala: How can I chain/nest implicit conversions?

原文：http://docs.scala-lang.org/tutorials/FAQ/chaining-implicits.html

 

The pimp my library pattern allows one to seemingly add a method to a class by making available an implicit conversion from that class to one that implements the method.

Scala does not allow two such implicit conversions taking place, however, so one cannot got from Ato C using an implicit A to B and another implicit B to C. Is there a way around this restriction?

Scala has a restriction on automatic conversions to add a method, which is that it won’t apply more than one conversion in trying to find methods. For example:

1. class A(val n: Int)
2. class B(val m: Int, val n: Int)
3. class C(val m: Int, val n: Int, val o: Int) {
4. def total = m + n + o
5. }
7. // This demonstrates implicit conversion chaining restrictions
8. object T1 { // to make it easy to test on REPL
9. implicit def toA(n: Int): A = new A(n)
10. implicit def aToB(a: A): B = new B(a.n, a.n)
11. implicit def bToC(b: B): C = new C(b.m, b.n, b.m + b.n)
13. // won't work
14. println(5.total)
15. println(new A(5).total)
17. // works
18. println(new B(5, 5).total)
19. println(new C(5, 5, 10).total)
20. }

However, if an implicit definition requires an implicit parameter itself, Scala will look for additional implicit values for as long as needed. Continuing from the last example:

1. // def m[A <% B](m: A) is the same thing as
2. // def m[A](m: A)(implicit ev: A => B)
4. object T2 {
5. implicit def toA(n: Int): A = new A(n)
6. implicit def aToB[A1 <% A](a: A1): B = new B(a.n, a.n)
7. implicit def bToC[B1 <% B](b: B1): C = new C(b.m, b.n, b.m + b.n)
9. // works
10. println(5.total)
11. println(new A(5).total)
12. println(new B(5, 5).total)
13. println(new C(5, 5, 10).total)
14. }

“Magic!”, you might say. Not so. Here is how the compiler would translate each one:

1. object T1Translated {
2. implicit def toA(n: Int): A = new A(n)
3. implicit def aToB(a: A): B = new B(a.n, a.n)
4. implicit def bToC(b: B): C = new C(b.m, b.n, b.m + b.n)
6. // Scala won't do this
7. println(bToC(aToB(toA(5))).total)
8. println(bToC(aToB(new A(5))).total)
10. // Just this
11. println(bToC(new B(5, 5)).total)
13. // No implicits required
14. println(new C(5, 5, 10).total)
15. }
17. object T2Translated {
18. implicit def toA(n: Int): A = new A(n)
19. implicit def aToB[A1 <% A](a: A1): B = new B(a.n, a.n)
20. implicit def bToC[B1 <% B](b: B1): C = new C(b.m, b.n, b.m + b.n)
22. // Scala does this
23. println(bToC(5)(x => aToB(x)(y => toA(y))).total)
24. println(bToC(new A(5))(x => aToB(x)(identity)).total)
25. println(bToC(new B(5, 5))(identity).total)
27. // no implicits required
28. println(new C(5, 5, 10).total)
29. }

So, while bToC is being used as an implicit conversion, aToB and toA are being passed asimplicit parameters, instead of being chained as implicit conversions.

See also:

• Context and view bounds
• A discussion on types, origin and precedence of implicits

This question and answer were originally submitted on Stack Overflow.