error[E0599]: no method named to_f32 found for struct euclid::Rect<Au, LayoutPixel> in the current scope

[kaiwk]

I want to convert au_rect: Rect<Au, LayoutPixel> to f32_rect: Rect<f32, LayoutPixel> by au_rect.to_f32(), as the compiler prompts:

26 | pub struct Au(pub i32);
   | ----------------------- doesn't satisfy `Au: num_traits::cast::NumCast`
   |
   = note: the method `to_f32` exists but the following trait bounds were not satisfied:
           `Au: num_traits::cast::NumCast`

Can we just impl NumCast for Au?

[emilio]

Hmm, do you expect the target rect to be in actual pixels I guess? (as in, the equivalent to to_f32_px()?)

I think NumCast can be confusing, because you may want the Au value as f32 (so value.0 as f32 or such), or the actual pixel value as f32 (value.as_f32_px())

[kaiwk]

as in, the equivalent to to_f32_px()?

Yes, exactly. It seems there is no way to convert the Rect<Au> directly.

because you may want the Au value as f32 (so value.0 as f32 or such)

ha, I see.

[mrobinson]

I think NumCast can be confusing, because you may want the Au value as f32 (so value.0 as f32 or such), or the actual pixel value as f32 (value.as_f32_px())

Out of curiosity, when is it useful to have value.0 as f32?