Scala pickle creates needless copies of objects about pickling HOT 8 CLOSED

I understand the relationship with #1 and the issue of the additional object being added to the pickle.

However, I don't understand how you're getting this behavior: X.b(0) == 4?

As far as I can see, it has to be X.b(0) == 1.

Example:

import scala.pickling._
import json._

object Main extends App {
  class X {
    var a = List(1,2,3)
    var b = a
  }

  val x = new X
  println(x.b(0))

  val p = x.pickle
  println(p.value)

  val u = p.unpickle[X]
  println(u.b(0))
}

Correctly Results in:

1
{
  "tpe": "Main.X",
  "a": {
    "tpe": "scala.collection.immutable.$colon$colon[scala.Int]",
    "elems": [
      1,
      2,
      3
    ]
  },
  "b": {
    "tpe": "scala.collection.immutable.$colon$colon[scala.Int]",
    "elems": [
      1,
      2,
      3
    ]
  }
}
1

That is b(0) is 1, not 4. Did I miss something? I'd just like to ensure something isn't semantically amiss with the pickle format.

from pickling.

It also seems that in the pickle you pasted into the issue description, you might have an array, not a list? Our JSONPickled output looks a bit different...

Even if I change X's fields to be an array instead of a list, I still get the correct expected output:

1
{
  "tpe": "Main.X",
  "a": [
    1,
    2,
    3
  ],
"b": [
  1,
  2,
  3
]
}
1

from pickling.

Yeah, I meant an array. The output contains the correct values, but the serialization should have some way of indicating that a and b are references to the same object, so that you don't get slightly different data when you serialize classes like these.

from pickling.

Yep, as mentioned I understand that– though you earlier mentioned some additional semantically incorrect behavior X.b(0) = 4. The issue description has since been edited, so I take it that all is OK, semantically, with accessing the 0th element of an unpickled instance of member b of X? That is, that point is a non-issue, right?

Good thing is that, as mentioned in related ticket #1, we're currently working to make pickling cyclic object graphs possible and more convenient– so both #1 and #2 should be resolved once our fix is pushed. :-)

from pickling.

Yep, as mentioned I understand that– though you earlier mentioned some additional semantically incorrect behavior X.b(0) = 4.

Yeah I meant to say what I'm now showing with the example code. Sorry I wasn't clear. Thanks a lot 👍

from pickling.

Ah, yeah totally missed that the example code changed too! Sorry! Alright, awesome, all clear now, thanks! :)

from pickling.

As commented on #1, this has been resolved in the "oopsla2013" branch (https://github.com/scala/pickling/tree/oopsla2013) and will be integrated into the main branches soon.

from pickling.

All of this is now in the main 2.10.x branch, so marking as fixed :)

from pickling.