Unable to use Self inside the struct definition for #[derive(Item)]

[Veetaha]

🐛 Bug description

Given this code:

use dynomite::Item;
use std::marker::PhantomData;

#[derive(Item, Clone)]
struct Book {
    #[dynomite(partition_key)]
    pub partition_key: Id<Self>,
    pub name: String,
}

trait MyTrait {}
impl MyTrait for Book {}

#[derive(PartialOrd, PartialEq, Clone, Debug)]
struct Id<T: MyTrait>(String, PhantomData<T>);

impl<T: MyTrait> dynomite::Attribute for Id<T> {
    fn into_attr(self: Self) -> dynomite::dynamodb::AttributeValue {
        todo!()
    }
    fn from_attr(value: dynomite::dynamodb::AttributeValue) -> Result<Self, dynomite::AttributeError> {
        todo!()
    }
}

but this code gives the following error:

Log

~/junk/rust-sandbox $ cargo check -p crate_foo
    Checking crate_foo v0.1.0 (/home/veetaha/junk/rust-sandbox/crate_foo)
error[E0277]: the trait bound `BookKey: MyTrait` is not satisfied
  --> crate_foo/src/main.rs:7:5
   |
7  |     pub partition_key: Id<Self>,
   |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `MyTrait` is not implemented for `BookKey`
...
15 | struct Id<T: MyTrait>(String, PhantomData<T>);
   |              ------- required by this bound in `Id`

error: aborting due to previous error

For more information about this error, try `rustc --explain E0277`.
error: could not compile `crate_foo`.

To learn more, run the command again with --verbose.
~/junk/rust-sandbox $ 

It looks like Id<Self> is copied into struct BookKey { ... } body but instead we should replace all Self occurences with the struct #[derive(Item)] is placed onto before copying it into BookKey.

🤔 Expected Behavior

This should work out of the box without compile errors.

The workaround is to use Book literally instead of Self

🌍 Your environment

dynomite version: 0.8.2 rustc version: rustc 1.44.1 (c7087fe00 2020-06-17)

[softprops]

It looks like Id<Self> is copied into struct BookKey { ... } body but instead we should replace all Self occurences with the struct #[derive(Item)] is placed onto before copying it into BookKey.

I need to parse this problem a bit. This sounds a bit like an edge case to me. I'm the majority of cases when you declare a field is a partition key the item key struct simply represents a copy of that field.

At the macro level there isn't really type information only syntax and I'm not sure how I would be able to infer something like this from syntax alone.

I'll think on this one.