JustForFun88

> Have you done some measurement that shows this? I would find an increase in compilation time here to be _very_ surprising Yes, my mistake, you are right 😊. Looks...

@panekj Made requested fixes and improved some of my solutions. I think this pull request is ready for another review.

> Where is the `CharBuffer` from (and why)? I wrote it myself as a helper structure to store a one-character `&str` on the stack instead of on the heap. I...

> Is there anything left to do? Yes, I need to clean up the PR by removing things that have already merged. Can't do it right now as I'm busy...

I would also suggest writing at least some documentation to the code. At least the documentation for the new code. I think examples of use can not be added. **But...

> I didn't notice #400 before, but I think that's doing a lot more than is really needed for the immediate problem. For instance, I don't think it's important to...

> I didn't notice #400 before, but I think that's doing a lot more than is really needed for the immediate problem. For instance, I don't think it's important to...

> I can rewrite `get_or_insert` and `get_or_insert_owned` to use `Entry` and `EntryRef` (with some minor changes). But for `get_or_insert_with` I would have to fall back to the `RawTable` API, which...

> That would be **unsound**, because a user could follow that with a `remove` call, making that reference dangling. That's why the raw entry has `fn into_key(self) -> &'a mut...

To be honest, the implementation of this, as well as the `get_or_insert_with` method, raises questions. These two methods make it quite easy to break the invariant of set that all...