Can someone clarify why reified types behave this ...

# announcements

[mp]

Can someone clarify why reified types behave this way?

Can someone clarify why reified types behave this way?

[Sam Garfinkel]

What is printed when you call println(ThingAndBoxedTypeRef.of<String>("a string").typeRef.type)

[Oscar]

And what if you do

inline fun <reified T> of(thing: T) =
    ThingAndBoxedTypeRef(thing, jacksonTypeRef<T>())

[araqnid]

I get something like Box<T> output for that

[araqnid]

it needs to be jacksonTypeRef<Box<T>>() to supply the explicit type parameter there

[araqnid]

It seems like the way to solve this is to go to the TypeFactory (Jackson stuff) and construct the type manually:

typeFactory.constructParametricType(Box::class.java, typeFactory.constructType(jacksonTypeRef<T>()))

Although this gives a JavaType rather than a TypeReference (which is just a type token way of making parameterised types easier to specify in source)

[araqnid]

but that doesn’t actually answer the question of why it ended up this way

[Sam Garfinkel]

I wonder if making a named subclass of TypeReference and instantiating that would make a difference instead of the ad-hoc anonymous object.

[Sam Garfinkel]

Since these are literally inlined there’s no reason we couldn’t simplify this by making this a single top level function composition, that might provide more insight.

[mp]

yep that pretty much summarizes my confusion: since it's all inlined it seems like it should all work out...

[mp]

Avoiding jacksonTypeRef by creating the object : TypeReference didn't help