Ah, we definitely don't want to generate an == that uses hash values! That's not correct.

Same questions as SR-7293: there was a reason we didn't add this for classes in the compiler; do we really want to add it as a refactoring? cc @allevato

Also, the hashing API is about to change, so I suggest people not take this yet until that's settled down. cc @lorentey

Indeed; I'm working on a proposal to overhaul Hashable. If it goes through, hashValue will be replaced with a function requirement.

An additional point on the template above: declaring hashValue as a stored property is almost never what we want – in the vast majority of cases, it should be declared as a calculated property.

var hashValue: Int {
  return <#combined value#>
}

The problem is, of course, that it's entirely non-obvious what expression people should write in the "combined value" slot. Hence the proposal I'm drafting!

Ok, I think i'm about to close this since relying on compiler to synthesize these conformances is a recommended strategy.

since we'll start to synthesizing the Hashable conformance, the refactoring may send the wrong message

Since `Hashable` is hard to get right, any code generated by a local refactoring either (1) would need to be nearly as good as what the compiler generates for value types, as a starting point for users to edit, or (2) would be extremely generalized to the point of being useless.

The catch is that the autogenerated `hashValue` for value types uses stdlib-private functions to provide a quality implementation, and a local refactor would not be able to do that. So item 1 above wouldn't be easy to achieve.

It's probably best left out.

@allevato I think you're right. The best strategy for the developers is to rely on compiler to synthesize these conformances. We should encourage them to avoid manually typing these boilerplate as much as we can. I've just closed this task.

I'm preparing a proposal to fix Hashable; if it passes, it will become possible to implement this refactoring well.

@nkcsgexi Do you remember what the resolution here was? Was the refactoring implemented or revoked in favor of derived conformances?