Сравнение D и C++ и Rust на примерах

10:02 SiegeLord: So I checked, D associative arrays in the current release treat NaN's as equal, and in the master they treat them as unequal (i.e. they do the bad thing)
10:02 SiegeLord: The bad thing being, I think you can insert key NaNs, but not get them back
10:03 SiegeLord: So… essentially… that code example is nothing to be proud of
10:04 kvark: SiegeLord: so it works in 99.9% of cases, where we don't have a NaN key, and the rest of the cases are simply unable to get the key back? That sounds like a win to me
10:04 SiegeLord: It's a DoS vector
10:04 pczarn: why?
10:05 SiegeLord: You make insertion operation be O(n)
10:05 SiegeLord: Keep inserting NaN, and it keeps making the bucket longer and longer
10:05 SiegeLord: And it has to check them all by equality before determining that its a 'new' key

10:09 pyon Why would anyone want to compare floats for equality? This is a data type invented for scientific computations with inexact numbers. :-|

use std::collections::hashmap::HashMap;
type MyMap = HashMap<String,int>;

fn a(mut map: MyMap) -> MyMap {
    map.insert("add".to_string(), 1);
    map
}

fn main() {
    let mut x: MyMap = HashMap::new();
    x.insert("o".to_string(), 2);
    println!("{}", a(x));
    x = HashMap::new();
    x.insert("c".to_string(), 4);
    println!("{}", a(x));
}

import std.stdio;

// Возврат из функции
int[string] add(int[string] array)
{
    array["add"] = 1;
    return array;
}

// Измененение будет видно вне функции
void add2(int[string] array)
{
    array["add"] = 1;
}

void main()
{
    writeln(add(["o": 2])); // ["add":1, "o":2]
    writeln(add(["c": 4])); // ["c":4, "add":1]
    
    // еще вариант записи
    writeln(["o": 2].add); // ["add":1, "o":2]
    // или даже так
    ["o": 2].add.writeln; // ["add":1, "o":2]
    
    // Демонстрация поведения by-reference
    int[string] arr = ["o": 2];
    add2(arr);
    writeln(arr); // ["add":1, "o":2]
}