- TODO:
- Figure out limitations to this approach. Will it scale with async for example
- Where should the __generated fn be placed? Is it good enough to put it in the module, or, especially for custom macro's, do we want the implementation to be stacked inside the #[maybe_my_custom_test] fn?
Can we replace the current binding based implementation completely? Or do we want or need to keep both?
#[parameterized(a = { 0 }, b = { 1 })]
fn numbers<T: Into<u8>>(input: T) {
assert_eq!(0, input)
}
mod test {
#[test]
fn a() {
fn __generated<T: Into<u8>>(input: T) {
assert_eq!(0, input.into());
}
__generated(0);
}
#[test]
fn b() {
fn __generated<T: Into<u8>>(input: T) {
assert_eq!(0, input.into());
}
__generated(1);
}
}
and
fn __generated(input: impl Into<u8>) {
assert_eq!(0, input.into());
}
__generated(0);
}
Can we replace the current binding based implementation completely? Or do we want or need to keep both?
and