HashCode of enum cases is not stable

[Lasering]

Compiler version

3.3.1

Minimized code

enum Test:
  case A

Test.A.hashCode

Scastie

Output

The outputted value changes for each run. Most likely the hashCode is being computed from the memory position.

Expectation

The hashCode should be stable across multiple JVMs.

[bishabosha]

so currently its the default java.lang.Object hashcode - this does make it a departure from case objects which singleton cases are meant to replace (when case objects are used as part of an ADT).

would this be considered a breaking change to make stable?

[bishabosha]

this isn't too wild to implement either, the shared anonymous class of the $new factory can derive the hashCode from the name parameter (do we want to only mix in name, or should ordinal count too?)

[sjrd]

Making it stable would not be a breaking change.

[Lasering]

This stackoverflow post suggests implementing the hashCode like this:

ordinal ^ this.getClass.getName.hashCode

because

would be deterministic across different JVMs. It would even work a bit better, since the least significant bits would "change as much as possible"

[odersky]

I don't see why Scala should be different from Java here. If you need a different hashcode it's very easy to compute it from oridinal.

[bishabosha]

The hashCode should be stable across multiple JVMs.

why is this a requirement? I would assume no one would use the hashcode as a persistent key