Generic types

[turion]

Fixes #424 . I'm trying to implement support for generic/polymorphic types. Can you give me some feedback whether I'm on the right track? If yes, I'll proceed with the other macros.

[evnu]

Can you give me some feedback whether I'm on the right track?

This looks like the right approach to me. Is something missing? How does it interact with lifetimes?

[turion]

Is something missing?

I don't know whether this feature is now already sufficiently tested, or whether I should write an extra test somewhere, and if yes what it should contain.

How does it interact with lifetimes?

So far only on the syntactical level. I didn't implement lifetime polymorphism here because I don't understand yet how they interact with the lifetimes of the Encoder and Decoder traits. The only reason I touched them a bit is because syntactically they are declared in the same place as type parameters.

[turion]

Is something missing?

Other derive macros are missing, e.g. for structs and enums.

[turion]

Ok, this should be good to go now. I added type parameters to all cases where it made sense (unit enums didn't make sense), and added appropriate tests I think.

[turion]

Ping

[evnu]

I like where this is going. If we could also lift the restriction on a single lifetime, that would be great.

[evnu]

A test with explicit lifetimes would be nice. Something along these lines maybe? Testing with lifetimes could be tricky, but a 'static string might suffice.

#[derive(NifStruct)]
struct Slice<'a> {
  my_str: &'a str
}

#[rustler::nif]
pub fn from_a_string_slice() -> Slice {
  Slice { my_str: "hello world" }
}

[turion]

Do you mean an explicit lifetime together with a type parameter? I (hope I) haven't changed anything about the behaviour with lifetimes here. I'll send a short PR ahead that adds a lifetime test (since it seems there isn't one yet in test_codegen.rs) and rebase onto that, then we know that I didn't break them.

[turion]

Ok, seems like lifetimes are broken anyways, #428. This looks like an orthogonal issue to me.

[turion]

Ping @evnu ;)

[evnu]

Ping @evnu ;)

Do you plan to complete #430 first and then we take a look into this PR here again?

[turion]

Do you plan to complete #430 first

Doesn't look like it. It's a too complicated issue, and orthogonal in content to this here. (But there will be merge conflicts)

[turion]

Clippy fails, but I didn't touch the failing code 😕 . The same with the formatter.

[evnu]

@turion can you rebase on master?

[turion]

@evnu The lifetimes MR has changed so much that this PR here can basically be thrown away. I'm not even sure, maybe generic types are already handled now?

[evnu]

I'm not even sure, maybe generic types are already handled now?

We will have to try how well adds generic types, but maybe we need to test this further. Lets keep this PR closed. For #424, we will have to check if that is actually fixed by #483.