Thibaut Goetghebuer-Planchon

The error seems unrelated (not an infinite loop) and more like a lack of memory when rehashing (hence the `std::bad_alloc` exception). How much RAM do you have?

It could be due to the `indices3` that are powers-of-two + the `std::hash` identity function that causes too much collisions and end-up to make the distance variable becomes too large...

Thanks. So I checked and added a try/catch on the `std::bad_alloc` to check the distribution of the values in the buckets (save the bucket for each value in a file...

Thank you, we effectively should avoid rehashing when no actual insert is done. I added some comments to the PR code.

Thank you for the change, looks good. Just a small comment left. Note that the map doesn't guarantee that erase is `noexcept` per the commit message. The problem is more...

Thanks, that would be a useful feature. I wonder though if there isn't a problem in the code where the last bucket is never cleared. If I do: ```c++ tsl::ordered_map...

Thanks for the update. It seems also that no backward shift is done on erasure (see the already existing erase functions in ordered-map or https://codecapsule.com/2013/11/17/robin-hood-hashing-backward-shift-deletion/). After erasing the elements, the...

Note also we may have to think about the strong exception guarantee of this function. What happen if an exception is thrown while we already have cleared some buckets? Would...

The ordered-hash only provides strong-exception guarantee if T is a noexcept movable-type or copyable one: > Concerning the strong exception guarantee, it holds only if ValueContainer::emplace_back has the strong exception...

Hi, I created a [branch](https://github.com/Tessil/hat-trie/tree/insert_with_prefix) to implement the feature. You can use it this way: ```c++ tsl::htrie_map map; map.insert_with_prefix("/home/user/", {{"test", 1}, {"test2", 2}, {"test3", 1}, {"test4", 2}, {"test5", 1}, {"test6",...