Josh Stone

@tesselode if you're interested, maybe you could comment on rust-lang/hashbrown#255 *why* you're concerned about this, and whether you think such a manual `vacuum` operation would work for you.

@malthe Sorry, I don't know how far you got, but I just pushed a rebase myself. I was responsible for a lot of the conflicting changes anyway, so I had...

To get some real data, I tried to use this branch in rust-lang/rust#96751 and its perf infrastructure. For [instructions], there were a few small wins, but nothing better than 1%...

I think ideally we would want two things for this, especially the first: 1. Stabilizing `HashMap::get_many_mut` so we can be sure to match that API as much as possible. 2....

> Is it okay to just leave the PR open until then? Yeah, that's fine. I should probably make a label for this...

I don't like using nightly/unstable features, in part because it's an extra maintenance burden, but also because the feature may be opted-in on your behalf, arbitrarily deep in your dependency...

Is there anything in this that will require crate-local access to `IndexMap`? I don't see anything so far. So I think having it in a separate crate will be fine,...

Hi, thanks for the update! Given that you have been publishing that crate for a while as well, I think we can close it here.

It's been this way ever since `PartialEq + Eq` were implemented for the original `OrderMap` in #23, with the goal of `HashMap` consistency, and I followed suit in #46 when...

I fear a proliferation of type parameters -- I proposed an index type in #147, and I think it's likely we'll want an [`Allocator`](https://doc.rust-lang.org/std/alloc/trait.Allocator.html) type once that stabilized in the...