Having:
def f () = {
(1, "two", 3.0)
}
Why is it ok
var (x, y, z) = f()
but not
var i = 0
var j = "hello"
var k = 0.0
// use i, j, k
...
//then
(i, j, k) = f() // ; expected but = found
?
Having:
def f () = {
(1, "two", 3.0)
}
Why is it ok
var (x, y, z) = f()
but not
var i = 0
var j = "hello"
var k = 0.0
// use i, j, k
...
//then
(i, j, k) = f() // ; expected but = found
?
You see here a limited version of pattern matching when initializing variables. Note that this works not only for tuples:
val a :: b = List(1,2,3)
println(a) //1
println(b) //List(2, 3)
This feature seems to be borrowed directly from Haskell, where you can use patterns for initialization as well:
let (a,b) = getTuple
in a*b
As Haskell has no mutable data, there is no way to assign something.
In Scala you could do something like this, but I guess this was considered too confusing, or maybe too difficult to implement. You can always use a match
expression as usual, and often you need just a case
, e.g. List((1,2),(3,4)).map{ case (a,b) => a*b }
.
My suspicion would be that without the "var" or "val" to the left of the tuple of variable names, the compiler treats the tuple as a tuple. That is, you're really trying to assign a value to an instance of Tuple3
and not to the three variables, and that makes no sense to the compiler.
Incidentally, using a function and various datatypes in your example isn't relevant. Here's a simpler example:
scala> var ( i, j, k ) = ( 1, 2, 3 )
i: Int = 1
j: Int = 2
k: Int = 3
scala> ( i, j, k ) = ( 4, 5, 6 )
<console>:1: error: ';' expected but '=' found.
( i, j, k ) = ( 4, 5, 6 )
^