Imported From: https://issues.scala-lang.org/browse/SI-6046?orig=1
Reporter: Michael Joya (mjml)
Affected Versions: 2.9.2
See #3448

@VladUreche said:
We need to have alpha-renaming in scaladoc to fix this systematically. OR, to fix this particular occurrence, we need to rename the Map type parameters to [K,V] for key and value, making the signature look like:

def map[B, That](f: ((K, V)) ⇒ B)(implicit bf: CanBuildFrom[Map[K, V], B, That]): That 

and the usecase:

def map[B, That](f: ((K, V)) ⇒ B): Map[B] // which is incorrect again, see SI-3448

What do you think?

And now the long explanation why this signature gets there:
Method map in scala.collection.Map comes from TraversableLike (which implements the declarations in FilterMonadic and GenTraversableLike):

trait TraversableLike[+A, +Repr] {
  ...
  def map[B, That](f: A => B)(implicit bf: CanBuildFrom[Repr, B, That]): That = {
    val b = bf(repr)
    b.sizeHint(this)
    for (x <- this) b += f(x)
    b.result
  }
  ...
}

Now, map is inherited into scala.collection.Map by way of:

trait Map[A, +B] extends Iterable[(A, B)] with ...
trait Iterable[+A] extends ... with IterableLike[A, Iterable[A]] 
trait IterableLike[+A, +Repr] extends ... with TraversableLike[A, Repr] with ...

But as you can see, Map extends Iterable[(A, B)], and thus substitutes A to ((Map's)A, (Map's)B). And now it's obvious how the signature is generated:

def map[B, That](f: ((Map's)A, (Map's)B) => B)(implicit bf: CanBuildFrom[Repr, B, That]): That = ...

@VladUreche said:
Alex, what would it take to alpha-rename all Map classes' type params from {{Map[A, B]}} to {{Map[K, V]}}? Would it mean touching 10 classes, 100 classes? Or would break other ppl's code?

@axel22 said:
Probably some 20-30 occurrences according to: grep -r "(trait|class) [A-Za-z]*Map[".

I don't think renaming the type parameters of Map classes should break any code.

@adriaanm said:
I vote against such a symptomatic fix. See #6057 for unexpected collateral damage of renaming type params

@VladUreche said:
Great! Let's aim for the rename then, as alpha-substitution in scaladoc has no way of getting into 2.10.
I'll have a look today and if I have trouble with the rename, I'll ask for help.

@VladUreche said:
Not sure I understand what's happening in #6057, but it looks like an error that should be fixed and that shouldn't stop us from renaming.

@axel22 said (edited on Jul 11, 2012 12:11:39 PM UTC):
Agreed with Adriaan here. First, it was a convention to use [A, B] as type parameters. Second, I think the risks of introducing breakage similar as in the linked ticket outweigh the gain of renaming..

@VladUreche said:
Hmpfff... Let me see what I can do about alpha-renaming then.