im-rs

They're using hash array mapped tries. I don't have my own personal implementation, I have been using https://github.com/bodil/im-rs until I can get around to making my own implementation (not that I really need to, but it would be a fun exercise). Functions generate a hash based on a unique id generated for the function, plus the hash of any captured variables, and a hash of the pointer address to the function). That is off the top of my head though so I could be missing some details. Hashing maps is tricky! With a sufficiently deep hash map you can run into problems since that invokes an equality check as well - at least how I handle it, is that you just attempt to naively hash the keys and values of the hash map, to create a hash code for that object. If the equality check ends up with a sufficiently large depth, eq returns false so we

Anecdotal evidence, but many large rust async code bases perform significantly better if you are using a single threaded runtime vs. a multithreaded runtime. Here is an issue for the quinn crate that implements QUIC: https://github.com/quinn-rs/quinn/issues/1433 We have had very similar experiences when developing https://github.com/n0-computer/iroh And this is with quinn still carrying around the synchronization primitives to make everything Send. Another piece of anecdotal evidence: the functional collection library https://github.com/bodil/im-rs comes in two flavours: one using Arc and one using Rc. The reason for this is that you get a significant performance increase by using Rc (where manipulating the refcount is just an integer inc/dec) vs Arc where manipulating the refcount needs to be an atomic op. Last but not least - often maki

Yay, it's no longer LGPL'd: https://github.com/bodil/im-rs/commit/32f58e914c84a6e965eea8... :)