Extension copy

[OndrejSpanel]

Started work on #234 by creating target test.

At the moment the test is failing with:

Unsupported path element. Path must have shape: .field1.field2.each. field3.(...), got: (($1: Vec) => x(_$1))

I am not sure how far I will get, I never tried to access extensions from macros, but hopefully it should be possible.

[OndrejSpanel]

I am familiarizing myself with the toPath method. It seems to me the extension should be processed in the branch case Apply(deep, idents). Actually no other code from tests goes through this path.

[OndrejSpanel]

First test passed: extension is found in a companion object of a plain class. This needed some adjustments in how is copy called in caseClassCopy, but nothing drastic (extension methods need to receive one more argument list). I will continue with different extension locations and then hopefully proceed to opaque types.

[OndrejSpanel]

More progress: I can use extensions to read source values as well.

There are few things I was struggling with. Most of them are related to the fact the functions are operating on symbols (Symbol), not types (TypeRepr), therefore finding suitable function overloads is sometimes a bit more messy than I would like.

The last step is to support opaque types. In first attempts I get assertions in the compiler when using them, hopefully I will be able to work through this somehow.

[OndrejSpanel]

I have hit a kind of obstacle: scala/scala3#22143

I am able to access the fields using extension methods, but if the underlying type already has fields with corresponding names, Symbol.fieldMember finds them and I assume they exist and can be used, which results in a bad resulting tree, with a.x instead of Vec.x(a), which later fails when compiling.

The whole Symbol / TypeRepr reflection is still something I do not understand much, therefore it is quite possible the behaviour I see is not a bug. No matter what, hopefully there will be some workaround, some way to find the a.x is not accessible.

[OndrejSpanel]

I think it should be done now.

[adamw]

@KacperFKorban maybe you could take a look as well?

[KacperFKorban]

@adamw Sure, I'll review it it on Monday.

[KacperFKorban]

I figured that it would be faster to just push some changes directly to the branch, hope you don't mind @OndrejSpanel
Looks good overall!

[KacperFKorban]

Is this still relevant? AFAIK this is just the modification function being called.

[KacperFKorban]

@OndrejSpanel This comment might've slipped through. Do you remember the reasoning behind this?

[OndrejSpanel]

I think it was just a comment made during exploration and it should be removed.

[KacperFKorban]

Ok, Thanks!

[OndrejSpanel]

hope you don't mind

Sure, better than requesting me to do them. I will have some questions regarding some changes.

[adamw]

Great, thanks! :)