Apply Expression on Input Argument

[mmstick]

There's some structures I'd like to use this with, but they require tiny adjustments to input arguments when storing them in the structure, such as subtracting 1 from an integer.

[Schyrsivochter]

Seconded. In one of my projects, I find myself wrapping things in a Box and cloning &strs to Strings a lot. It’d be nice if there was a way to specify this in a derive attribute. Something like:

#[derive(new)]
pub struct Foo {
    #[new(eval = "Box::new(bar)")] bar: Box<Bar>,
    baz: Baz,
}

The exact keyword is up for debate: eval, expr, apply, convert etc. Also, if this is even possible: instead of repeating the field name, a shorthand could be introduced to represent the input parameter for the field, e.g. it or <> or _.

[wycats]

A common use-case for me along these lines is making the argument impl Into<T> and the field T. This feature would enable that use-case, but I wonder if a dedicated attribute for this use-case would make sense:

#[derive(new)]
pub struct Person {
  #[new(into)]
  name: String;
}

Another common use-case for me that this feature would help with, but which might also justify dedicated syntax is to clone the input expression.

#[derive(new)]
pub struct Shared {
  #[new(cloned)]
  write: Arc<Something>,
}

[kriswuollett]

The addition of into and into_iter in #43 has been helpful for singular fields.

Not sure about custom expressions, but also a dedicated owned could support Person::new("John", ["Mary", "Beth"]) for the following?

#[derive(new)]
pub struct Person {
  name: String,
  #[new(into_iter="&str", owned)]
  friends: Vec<String>,
}

A specific use case where this optimization would help a lot is constructing resources in kube-rs when there are Vec<String> fields.