How can I use a custom hash function in a HashSet or HashMap?

Question

Since SipHasher is too slow for my use case, I'm trying to implement a custom hash function. I found an example which I used as base to get everything compiling.

My current code looks like this:

use std::collections::hash_state::{DefaultState};
use std::collections::{HashMap, HashSet};
use std::default::Default;
use std::hash::{Hash, Hasher, SipHasher};
use std::marker;

pub struct FnvHasher(u64);

impl Default for FnvHasher {
    fn default() -> FnvHasher { FnvHasher(0xcbf29ce484222325) }
}

impl Hasher for FnvHasher {
    fn write(&mut self, bytes: &[u8]) {
        let FnvHasher(mut hash) = *self;
        for byte in bytes {
            hash = hash ^ (*byte as u64);
            hash = hash * 0x100000001b3;
        }
        *self = FnvHasher(hash);
    }
    fn finish(&self) -> u64 { self.0 }
}

fn main() {
    let mut set:HashSet<i64, DefaultState<FnvHasher>> = HashSet::with_hash_state(DefaultState::<FnvHasher>);
}

When I compile I get the following error message:

$ rustc -V
rustc 1.0.0-nightly (522d09dfe 2015-02-19) (built 2015-02-19)

$ rustc hash.rs
hash.rs:26:86: 26:111 error: mismatched types:
 expected `std::collections::hash::state::DefaultState<FnvHasher>`,
    found `fn(core::marker::PhantomData<FnvHasher>) -> std::collections::hash::state::DefaultState<FnvHasher> {std::collections::hash::state::DefaultState}`
(expected struct `std::collections::hash::state::DefaultState`,
    found fn item) [E0308]
hash.rs:26         let mut set:HashSet<i64, DefaultState<FnvHasher>> = HashSet::with_hash_state(DefaultState::<FnvHasher>);
                                                                                                ^~~~~~~~~~~~~~~~~~~~~~~~~
error: aborting due to previous error

How can I get the above example to compile or use a custom hash function in a HashSet/HashMap?

score 6 · Answer 1 · answered May 05 '16 at 20:43

Rust 1.7 stabilized the use of custom hashers. For example, you can use FnvHasher from the fnv crate

extern crate fnv;

use std::collections::HashMap;
use std::hash::BuildHasherDefault;
use fnv::FnvHasher;

type HashMapFnv<K, V> = HashMap<K, V, BuildHasherDefault<FnvHasher>>;

fn main() {
    let mut map = HashMapFnv::default();
    map.insert(1, "Hello");
    map.insert(2, ", world!");
    println!("{:?}", map);
}

score 5 · Accepted Answer · answered Feb 21 '15 at 10:04

Take a look at the definition of DefaultState:

pub struct DefaultState<H>(marker::PhantomData<H>);

You are treating it as though it were as it used to be:

pub struct DefaultState<H>;

This change is recent; generic parameters must now be expressly used.

For this you must now use Default::default(), for that field is private, preventing literal expressions from being written. Thus, DefaultState::<FnvHasher> becomes Default::default().

And once you’re using Default::default() for that, you might as well go the whole hog and replace HashSet::with_hash_state(Default::default()) with Default::default():

let mut set: HashSet<i64, DefaultState<FnvHasher>> = Default::default();

How can I use a custom hash function in a HashSet or HashMap?

2 Answers2

Linked