Subclasses of `TypeName` return `TypeName` from `copy`

[Vampire]

If you for example call copy on a ClassName, you get TypeName returned.
The copy with tags parameter is overridden in the subclasses and returns the respective subclass.
But the copy without tags parameter always returns TypeName.

It would be nice if TypeName could get a type parameter that the subclass sets to itself and then both copy functions return the actual subclass.

[Egorand]

Seems like that would break binary compatibility, so we'd have to postpone the fix until 2.0.

[Vampire]

Are you sure it breaks binary backwards compatibility?
I remember a similar case where in the JVM stdlib ByteBuffer.flip was changed to be overwritten and return ByteBuffer in Java 9 when before it was inherited from the superclass and thus returned Buffer.
If you now compile ByteBuffer.allocate(2).flip(); with Java 8 it runs on Java 8 and Java 11.
If you instead compile with Java 11 (with source and target set to 8), it runs on Java 11 but throws java.lang.NoSuchMethodError: java.nio.ByteBuffer.flip()Ljava/nio/ByteBuffer; on Java 8.

Derived from that, I'd say it is a backwards compatible change but not a forward compatible one.
So as long you do not support compiling against 1.19.0 while running against 1.18.1, I'd say it should be fine.

This should of course be tested, and I'm not in a rush as I have a work-around.
Just wanted to mention that I think this might not be breaking.
Not more than adding a new method for example.

[ZacSweers]

It should be binary-compatible to specify a type, it would be an issue the other way around

[Egorand]

Makes sense, thanks! Guess yeah, we should be fine since we'll be changing the return type to subtype.