scottmcm

https://github.com/rust-lang/rust/issues/87991 is a good example of the kind of accident that having a feature like this would help prevent.

> based on the belief that core should attempt to never panic This is an overstatement. There's even panics in *built-ins*, like `[1][1]`. That doesn't mean these APIs *should* panic,...

`with_position_2` would make me sad. Maybe an inherent method for `Position -> (Position, T)`? So you could always do `.with_position().map(Position::into_parts)` or similar for the cases where the tuple is more...

Thanks for the PR! While I'm sometimes willing to add new APIs without libs-api assent, I'm much more worried about churning signatures without contacting them to see if they like...

Quick demonstration that .0 is rounding odd while the other numbers of decimal places are rounding even: ``` [src/main.rs:2] format!("{:.0}", 0.5) = "1" [src/main.rs:3] format!("{:.1}", 0.25) = "0.2" [src/main.rs:4] format!("{:.1}",...

Oh! I hadn't realized it was *just* `0.5` with the problem. That pushes me to "this is just a bug", not really a decision to be made. (Still worth a...

> `[src/main.rs:9] format!("{:.1}", 0.35) = "0.3"` > `[src/main.rs:10] format!("{:.2}", 0.155) = "0.15"` > `[src/main.rs:11] format!("{:.2}", 0.165) = "0.17"` Remember that anything that's not a multiple of an integer power of...

@blueglyph Note that floating-point algorithms are expected to treat the floating point number as if the represented value was exactly the desired value, and not to attempt to divine whether...

> but I'm not convinced it's true for their _decimal representation_, since we know the base 2 representation is not exactly the desired value, as we saw earlier. I'm not...

@bors r+