Incorrect handling of Option[xxx] if None

[jcrowley66]

Apologies if this is a duplicate, but several searches did not turn up this problem. [Using JSON format]

With this definition: case class PickleNone(test:Option[String])

this sequence works OK: val test = PickleNone(None) val roundTrip = test.pickle.value.unpickle[PickleNone]

gives test == roundTrip

but this sequence: val test = PickleNone(None) val roundTrip = test.pickle.value.unpickle[AnyRef]

results in test != roundTrip

It appears in the second case that the unpickle step sees the scala.None.type and instantiates a new instance for None. Unfortunately, Scala expects None to be a singleton, and any None == None test is reduced to object identity, which now fails.

Scala 2.11.8 Pickling 0.10.1

Full test case follows

package pickletests

import scala.pickling._ import scala.pickling.Defaults._ import scala.pickling.json._

case class PickleNone(testOption:Option[String])

object PickleNoneTest extends App { val test = PickleNone(None) val testPickled = test.pickle.value val testUnpickled = testPickled.unpickle[PickleNone] val testAnyRef = testPickled.unpickle[AnyRef] val testAnyRefAsPickleNone = testPickled.unpickle[AnyRef].asInstanceOf[PickleNone]

Console.println("test == testUnpickled is: " + (test == testUnpickled)) Console.println(" test == testAnyRef is: " + (test == testAnyRef)) Console.println(" test == asInstanceOf is: " + (test == testAnyRefAsPickleNone))

Console.println("") Console.println("") Console.println(" test.testOption is: " + test.testOption.toString + " @%8X".format(System.identityHashCode(test.testOption))) Console.println(" testUnpickled.testOption is: " + testUnpickled.testOption.toString + " @%8X".format(System.identityHashCode(testUnpickled.testOption))) Console.println(" testAnyRef.testOption is: " + testAnyRef.asInstanceOf[PickleNone].testOption.toString + " @%8X".format(System.identityHashCode(testAnyRef.asInstanceOf[PickleNone].testOption))) Console.println("testAnyRefAsPickleNone.testOption is: " + testAnyRefAsPickleNone.testOption.toString + " @%8X".format(System.identityHashCode(testAnyRefAsPickleNone.testOption))) }

[michielboekhoff]

Just tested this, this seems to have been fixed in 0.11.x. Can you confirm this, @jcrowley66?

[jcrowley66]

Ran under both 0.10.1 and 0.11.0-M2. JDK 1.8.0_121 in both cases. Both failed, and 0.11.0-M2 seems to have regressed, since each of the None instances now has a separate object ID. (Sorry about the formatting - can't quite figure it out!)

"C:\Program Files\Java\jdk1.8.0_121\bin\java"

Using scala pickling 0.10.1

test == testUnpickled is: true test == testAnyRef is: false test == asInstanceOf is: false

                  test.testOption is: None @4940809C
         testUnpickled.testOption is: None @4940809C
            testAnyRef.testOption is: None @7A138FC5
testAnyRefAsPickleNone.testOption is: None @379AB47B

Process finished with exit code 0

Using scala pickling 0.11.0-M2

test == testUnpickled is: false test == testAnyRef is: false test == asInstanceOf is: false

                  test.testOption is: None @6928F576
         testUnpickled.testOption is: None @ 9CD25FF
            testAnyRef.testOption is: None @27E0F2F5
testAnyRefAsPickleNone.testOption is: None @3574E198

Process finished with exit code 0

[jcrowley66]

Sorry re the Close - hit the wrong button. Both of the previous tests run under Windows 7.

[jcrowley66]

Set up a completely clean project (using IntelliJ) with PickleNoneTest as the only source code present. Ran with 0.10.1 as the only Jar included, then deleted that and ran with 0.11.0-M2 as the only Jar. Same results.

[michielboekhoff]

I'll have a look at this later today.

[michielboekhoff]

I've been able to replicate this issue. Interesting that I hadn't been able to replicate this. I'll have a look to see if I can find a fix.

[jcrowley66]

Great, thanks.

John Crowley Westport, CT 203-856-2396

On Jun 7, 2017, at 10:43 AM, Michiel [email protected] wrote:

I've been able to replicate this issue. Interesting that I hadn't been able to replicate this. I'll have a look to see if I can find a fix.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/scala/pickling/issues/453#issuecomment-306816891, or mute the thread https://github.com/notifications/unsubscribe-auth/Aamnt3zCA47raIIRujGfD3W1SKbhU6n1ks5sBrcbgaJpZM4NrqYh.

[michielboekhoff]

Right, so this issue is intermittently showing. I can't figure out exactly what it is. Anyone available to provide some assistance?

[michielboekhoff]

Right, I've done some more investigation and I'm going to share my findings:

Sometimes, the test I've made passes, seemingly randomly. I think what's happening is that a generated pickler for None is creating a new instance, which gets assigned a different ID. As this is not intended, I would think, it's best to write a manual pickler for Option, which I can have a go at tomorrow.

[jcrowley66]

Agree, when scala.None.type is detected while unpickling it needs to be handled separately so that the unique scala.None instance is returned. Sorry that I have no experience with the innards of the pickling library, so cannot offer any advice on the implementation.

[michielboekhoff]

I don't think that'd be too hard, I'll set it up. Just wondering why the test passes intermittently... maybe force either runtime pickling or static pickling, see if that makes it better?

[jcrowley66]

Now you are definitely beyond my expertise!

[jcrowley66]

This might be a hint - slightly different problem, but in the same ballpark. This is 0.10.1 - will try to run with 0.11.x tomorrow & report if any difference.

This source: responseTo: Option[UUIDCarrier] = None, // None if this is original message gets pickled as: "responseTo": { "$type": "scala.None$" },

which then throws an exception while trying to unpickle:

scala.pickling.PicklingException: Tag scala.None$ not recognized, looking for one of: scala.None.type, scala.Some[realtime.edt.common.UUIDCarrier]