Olivier Giniaux

We might have to bump to `3.1.2` and re-publish as well, no?

This turns out to be much more challenging than expected: the target_feature disables inlining, which defeats the purpose of an fast hashing algorithm. This is done by design for "safety...

Interesting, so if I understand correctly, with your changes, the `Hasher` remains even in no `std` builds, but with the following subtleties: - When `std` feature, it creates a seed...

Nice! I have just cherry-picked your changes. Next step is to have this covered by tests/CI and add a few lines about it in the readme. I can do it...

Hello! I'm sorry I haven't provided any update here. The `no_std` is still very relevant. Since there might be some changes regarding the `Hasher` construction and the dependencies, I've cherry-picked...

There is an [meowhash](https://github.com/bodil/meowhash-rs/tree/master) crate but it seems half-baked (wasn't even able to build with it on Macbook ARM, so I doubt it will even be correct or perform as...

Indeed, that would probably increase performance in scenarios where a `BuildHasher` is created for hashing only a few bytes. This RandomState w/ increment is interesting (and battle-tested since it's in...

From what I can see there is a compiler optimization that stack allocates `[0i8; VECTOR_SIZE]` instead of heap allocating (probably because `VECTOR_SIZE` is a constant), so `MaybeUninit` may not be...

About to close this one unless someone has some snippet to propose?

Closing this as proposed solution does not provides significant performance gains nor simplifies the code. Feel free to open another issue if you have something to suggest.