SICP Exercise 1.3: Sum of squares of two largest numbers out of three, Rust Version

Question

The exercise 1.3 of the book Structure and Interpretation of Computer Programs asks the following:

Exercise 1.3. Define a procedure that takes three numbers as arguments and returns the sum of the squares of the two larger numbers.

My answer is this:

#![feature(core)]
use std::cmp;
fn sum_square_largest(x:f64, y:f64, z:f64) -> f64 {
 match partial_min_three(x, y, z) {
 Some(a) => x * x + y * y + z * z - a * a,
 None => 0.0_f64/0.0_f64
 }
}
fn partial_min_three<T>(v1: T, v2: T, v3: T) -> Option<T> where T: PartialOrd<T> {
 match cmp::partial_min(v2, v3) {
 Some(x) => cmp::partial_min(v1, x),
 None => None
 }
}

Rust is the language I know less, I would really appreciate your advice.

Answer 1

I find it always pays to be tidy and conscientious about spacing. Here, I'd add spaces after the : in the argument declaration

// fn sum_square_largest(x:f64, y:f64, z:f64) -> f64 {
fn sum_square_largest(x: f64, y: f64, z: f64) -> f64 {

Next, I'd wrap the where clause onto the next line. where clauses often get large, so I find it best to let them have breathing room, and only put one type per line:

// fn partial_min_three<T>(v1: T, v2: T, v3: T) -> Option<T> where T: PartialOrd<T> {
fn partial_min_three<T>(v1: T, v2: T, v3: T) -> Option<T>
 where T: PartialOrd<T>
{

Checking for Some or None and doing something is a common pattern. In this case, you can use Option::and_then:

// match partial_min(v2, v3) {
// Some(x) => partial_min(v1, x),
// None => None
// }
partial_min(v2, v3).and_then(|x| partial_min(v1, x))

Programming is often about clarity in expressing your intentions, and I doubt that the majority of programmers immediately know what 0 / 0 is for floating point numbers. You should be explicit:

// None => 0.0_f64/0.0_f64
None => std::f64::NAN,

Again, we can use existing patterns to deal with checking for Some / None. This time, we can use Option::map:

// match partial_min_three(x, y, z) {
// Some(a) => x * x + y * y + z * z - a * a,
// None => std::f64::NAN,
// } 
partial_min_three(x, y, z)
 .map(|a| x * x + y * y + z * z - a * a)
 .unwrap_or(std::f64::NAN)

I disagree with your choice of NaN as a magic value. Magic values drive me crazy, and Rust has great choices for avoiding them - Option and Result!

fn sum_square_largest(x: f64, y: f64, z: f64) -> Option<f64> {
 partial_min_three(x, y, z)
 .map(|a| x * x + y * y + z * z - a * a)
}

I'd probably create a tiny function for squaring. It's likely to get inlined, so I'm not worried about performance, just readability:

.map(|a| square(x) + square(y) + square(z) - square(a))

All together (playpen):

fn sum_square_largest(x: f64, y: f64, z: f64) -> Option<f64> {
 partial_min_three(x, y, z)
 .map(|a| square(x) + square(y) + square(z) - square(a))
}
fn partial_min_three<T>(v1: T, v2: T, v3: T) -> Option<T>
 where T: PartialOrd<T>
{
 partial_min(v2, v3).and_then(|x| partial_min(v1, x))
}
fn square(x: f64) -> f64 { x * x }
// Copied from the standard library as it is currently unstable
// and the beta doesn't allow unstable features
use std::cmp::Ordering;
fn partial_min<T: PartialOrd>(v1: T, v2: T) -> Option<T> {
 match v1.partial_cmp(&v2) {
 Some(Ordering::Less) | Some(Ordering::Equal) => Some(v1),
 Some(Ordering::Greater) => Some(v2),
 None => None
 }
}
fn main() {
 let z = sum_square_largest(1.0, 2.0, 3.0);
 println!("{:?}", z);
}