What's the rationale for this suggestion?

similar to the suggestion for updating

class

to

data class

I guess

similar to the suggestion for updating

class

to

data class

I guess

I don't get such suggestion. It seems the object one is related to the fact that it's inside a sealed interface

Perhaps it's related to this recent discussion. Sealed objects are often used in exhaustive

when

clauses, but if the object comes from Java serialization, or from a mocking framework, the object may not be an actual singleton and the

==

operator used by the

when

clause may return false even for objects of the same type. Making it a

data object

fixes this possibility.

hmm interesting!

Also, the default

toString()

of a non-data object is platform-dependent while

toString()

of a data class is not. This can be problematic in KMP apps for things like logging, when the platform’s representation doesn’t include any type information. For example, see how the

toString()

here changes between JVM and JS targets (and how completely useless the JS version is) https://pl.kotl.in/OZ8JtYfeC

I'm now wondering why

data object

exists and

object

is not doing "the right thing" by default 🙂 But I suppose I'm not seeing the full picture (maybe that would be a breaking change...)

Many top-level objects don’t need all the extra stuff generated by the

data object

, and it would increase the binary size significantly to make them all data objects. A normal object is simple a singleton class with no extra ceremony. It’s just that the common semantics of an object inside a sealed class is closer to that of a data class that doesn’t have members, rather than an actual singleton, hence the addition of

data object

and its Intellij inspection

that makes a lot of sense! Thanks 🙏