Set equality is unexpectedly ordered

[rnewman]

  it('Compares sets correctly.', (done) => {
    expect(new Set(['a', 'b'])).toEqual(new Set(['b', 'a']));
    done();
  });

      Error: Expected Set (2) {'a', 'b'} to equal Set (2) {'b', 'a'}
      + expected - actual

      -["a","b"]
      +["b","a"]

This is with 1.18.0. I expect set equality to be independent of order; the order of elements is an implementation detail.

[ljharb]

Your expectation is incorrect - Set in JavaScript has a defined and consistent implementation order, and is a spec detail, not an implementation detail.

[rnewman]

I wouldn't expect the preserved insertion order of the set — which is noted in the spec — to influence its equality properties, but hey, JS is full of surprises. Time to switch to Immutable.js for this, I guess.

[ljharb]

I think it'd be reasonable to have a method that explicitly checked that two Sets/Maps/arrays had the same contents, irrespective of order.

[mjackson]

@ljharb What would you call the to method?

[ljharb]

hmm - naming is hard :-) something like expect(foo).toBeTheSameSetAs(bar) would work but that uses the math definition of Set, not the JS one. expect(foo).toMatch(bar) maybe?

[mjackson]

How about expect(a).toEqualSet(b)? expect-jsx already uses toEqualJSX.

[mjackson]

Actually, just found this:

Two sets A and B are said to be equal if every element of A is an element of B and every element of B is an element of A. The equality of sets A and B is denoted by A = B. Inequality of two sets A and B is denoted by A ≠ B.

Seems like we should just do some type detection in toEqual and toNotEqual and compare sets by their contents, irrespective of order.

[ljharb]

@mjackson if it's generic, that would break tests that currently compare array orderings. If it's not - and only for Set/Map - then you only have a slow try/catch way of brand-checking that it's a Set or a Map, and, Sets and Maps in JS are implicitly ordered.

[mjackson]

Even though they are implicitly ordered, conceptually a Set is not. toEqual is the right method IMO.

[ljharb]

I don't agree in the case of JS - but it wouldn't be a tragedy imo to have toEqual compare Sets and Maps in an unordered fashion - if i want ordering, I can [...foo] and compare those.

[mjackson]

I'd be open to a PR that does this, @rnewman, if you're up for it. :)