tison

@programmerjake Yeah. Then if we follow the Hash/Hasher/BuildeHasher analogy, we may have: * Compare/BuildCompare * PartialCompare/BuildPartialCompare * CompareEq/BuildCompareEq * ~PartialEq~ (UPDATE: I noticed that Eq and PartialEq have the same...

> they have significant semantic differences, so I think we need both ComparePartialEq and CompareEq. Well. Then maybe for naming: * Compare * ComparePartial * CompareEq * ComparePartialEq > Default...

I've updated the description "Solution sketch" for adding four traits (analogy to PartialEq/Eq/PartialOrd/Ord). For collections like `BinaryHeap` and `BTreeMap`, we can add a new state field and its type parameter,...

For HashMap, I may need some time to work a clear solution. But it seems the internal HashMap(HashTable) can have a comparator as state to wrap the `equivalent` implementation?

> An alternative approach would be a separate type along the lines of hashbrown's [HashTable](https://docs.rs/hashbrown/latest/hashbrown/hash_table/struct.HashTable.html) which gives you full control over hashing and equality comparisons. An advantage of this approach...

> Can you explain how just having the traits in core without further integration into the rest of the standard library would improve the problems you're facing? You can check...

> My consideration is that, if a `Comapre` trait is suitable in the core lib while other collection would benefit from it, we can first introduce the trait so that...

I found: https://github.com/smol-rs/futures-lite/blob/445f3fc922bea7993713f0f80b9efa96b152f7ce/FEATURES.md?plain=1#L187 .. that describes a similar combinator in futures-lite. How should I set expectations between futures-rs and futures-lite? Especially given that (1) futures-rs is under the rust-lang org...

I make a new type wrapper as a workaround: https://github.com/andylokandy/nom-rule/pull/8/files/62e78a4a0475eba6494b49820e91b414474e432e..6fcd576dc7022955f90dc837476d5907331ad621 But it's not so ergonomic, yes. > I think there should be an implementation for Input for all slices of...

@marcdejonge That should help. Thank you! We're using a wrapper `Input` type inside our database application, which provide the Span and Backtrace during parsing to provide located error reports. Perhaps...