Make ksuid hashable so it can be used within HashSet & HashMap

[aurelien-clu]

Feature Request

Motivation

I would like to use ksuid within an HashMap<Ksuid/KsuidMs, T> and I would prefer not to convert to a string, i.e. HashMap<String, T>.

Right now I am wrapping it within a struct (I am not aware of a better way, like a macro to implicitly call to_string before hashing):

use std::hash::{Hash, Hasher};
use svix_ksuid::Ksuid;

pub struct Id(pub Ksuid);

impl Hash for Id {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.0.to_string().hash(state);
    }
}

Proposal

I believe the following would be enough:

use std::hash::{Hash, Hasher};

impl Hash for Ksuid {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.to_string().hash(state);
    }
}

impl Hash for KsuidMs {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.to_string().hash(state);
    }
}

Pros:

• simple
• not prone to bugs or hash collisions (not more than ksuid as far as my understanding)

Alternatives

Having an intermediate String before computing the hash is not the most efficient way to do this but I personally would prefer to have something even if it is not the most performant 😇

Next

If you don't mind my proposal, would you like me to make a PR with it?

If you do mind adding this, fair enough, I will handle this on my end. 😉

If you do mind using the intermediate string, do you already have in mind a way to do it or would you like me to look into it? (my experience is limited)

---

Either way, thank you for the library 😄

[tasn]

Thanks a lot, and yes, please open a PR!

Is it possible to use the internal bytes representation though? To save us from going through a to_string?

[aurelien-clu]

Sure, I have just checked your source code, the internal representation is only:

pub struct Ksuid([u8; TOTAL_BYTES]);

I thought you had both timestamp and byte array and that it could be cumbersome.

I will probably do this tomorrow. (it is late right now, where I live) 😉

ps: thank you for answering quickly :)

[tasn]

Yup, the internal representation is just the byte array, so it should be easy. :)

Looking forward to seeing that PR!