Crate ord_subset [] [src]

Ever wanted to call .max() on an iterator of floats? Now you can! Well, almost: .ord_subset_max().

This crate is for types like the float primitive types f32 and f64: Types that are totally ordered except for these particular values.

I call these types subset-ordered. They can be marked with the OrdSubset trait that this crate defines. Such types can be put in the OrdVar struct. Wrapping your value in this marks to other code that the contents are ordered, thus fulfilling generic Ord trait bounds.

For convenience, iterators and slices are extended so that OrdSubset types have access to methods equivalent to .max() and .sort(). Values in the unordered subset of a type that is OrdSubset are handled in a consistent manner (Ignored or put at the end).


Add this to your Cargo.toml:

ord_subset = "2"
use ord_subset::{OrdSubsetIterExt, OrdSubsetSliceExt};

// Slices. Works on vector, too.
let mut s = [5.0, std::f64::NAN, 3.0, 2.0];
assert_eq!(&s[0..3], &[2.0, 3.0, 5.0]);
assert_eq!(s.ord_subset_binary_search(&5.0), Ok(2));

// iterators
assert_eq!( s.iter().ord_subset_max(), Some(&5.0) );
assert_eq!( s.iter().ord_subset_min(), Some(&2.0) );


Licensed under the Apache License, Version 2.0 or the MIT license, at your option. This file may not be copied, modified, or distributed except according to those terms.



Wrapper to signal that the contained variables have a total order. It's illegal to compare two OrdVars that are not ordered. For this reason, it's unsafe to create OrdVars without checking. Checked constructors are available for OrdSubset types.



Trait for types that form a total order when a few values are disallowed.