Rustc can't resolve type equality when given in where clause

help
1

I'm writing an algebraic numbers library, and I've run into a bug where rustc gives only one error:

error[E0284]: type annotations required: cannot resolve `for<'a> <T as std::ops::Div<&'a T>>::Output == T`
   --> src/polynomial.rs:602:32
    |
602 | impl<T: PolynomialCoefficient> PolynomialDivSupported for T
    |                                ^^^^^^^^^^^^^^^^^^^^^^

The relevant code:
polynomial.rs:602-610

impl<T: PolynomialCoefficient> PolynomialDivSupported for T
where
    T: for<'a> ExactDiv<&'a T, Output = T>
        + for<'a> Div<&'a T, Output = T>
        + AlwaysExactDivAssign
        + for<'a> AlwaysExactDivAssign<&'a T>,
    for<'b> T::Element: ExactDivAssign<&'b T::Element> + One + ExactDivAssign,
{
}

I'm pretty sure I missing trait bounds somewhere and/or there's a rustc bug. Any suggestions?

2

This Output throws it off

pub trait AlwaysExactDiv<Rhs = Self>:
    ExactDiv<Rhs> + Div<Rhs/*, Output = <Self as ExactDiv<Rhs>>::Output*/>
1 Like
3

I would have expected rustc to be able to resolve that since AlwaysExactDivAssign requires Output = Self:
traits.rs:580-587

pub trait AlwaysExactDivAssign<Rhs = Self>:
    AlwaysExactDiv<Rhs>
    + ExactDivAssign<Rhs>
    + DivAssign<Rhs>
    + ExactDiv<Rhs, Output = Self>
    + Div<Rhs, Output = Self>
{
}

I'll try removing the bound in AlwaysExactDiv and get back to you.

4

your solution does work, thanks!

5

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.