[smasher164]

Another approach is to use a Gc<T> smart pointer: https://docs.rs/gc/latest/gc/

I've used this in an interpreter and it's quite convenient.

[zozbot234]

Yes, garbage collection seems to be the only viable solution for dealing with spaghetti reference graphs in their full generality, including possible cycles. In that context it's worth trying a high-performance concurrent GC implementation such as https://github.com/chc4/samsara Samsara.

[celeritascelery]

That is just using a RefCell under the hood[1] so it is effectively the same trade-offs as the RefCell example from the article.

[1]https://docs.rs/gc/0.5.0/src/gc/lib.rs.html#495-498

[rendaw]

It doesn't require you to differentiate weak and strong references, so there's no risk of memory leaks due incorrect choice in Gc, unlike Rc IIUC.

[paholg]

Note: the two of you are linking to different crates. Samsara provides a Gc type, but it is not the gc crate.

[foldr]

Nice! This seems way more appealing than every other approach I've seen in this domain.