[@types/sinon] sinon.spy lacks the third argument

[karak]

If you know how to fix the issue, make a pull request instead.

• I tried using the @types/sinon package and had problems.
• I tried using the latest stable version of tsc. https://www.npmjs.com/package/typescript
• I have a question that is inappropriate for StackOverflow. (Please ask any appropriate questions there).
• Mention the authors (see Definitions by: in index.d.ts) so they can respond.
  • Authors: @MrBigDog2U @rationull @lumaxis @nicojs @43081j @tdreyno @JoshuaKGoldberg

If you do not mention the authors the issue will be ignored.

We can't spy accessors because the current definition of sinon.spy lacks third argument called types, which should be typed as types?: string[].

See https://github.com/sinonjs/sinon/blob/master/lib/sinon/spy.js.

My test project is here: https://codesandbox.io/s/92n50m2n1p.

[nicojs]

Is this an undocumented feature? I don't see it here: https://sinonjs.org/releases/v7.2.3/spies/#creating-spies-sinonspy-method-signatures

Generally, we only add typings of features that are documented.

[43081j]

It is undocumented.

I suggest opening an issue with sinon to find out if it's supported (as in they won't just remove it one day and it isn't deprecated).

[karak]

OK. I issued.

[karak]

It's supported though still undocumented.
sinonjs/sinon#1970 (comment)

[nicojs]

Cool, feel free to create a PR (for both sinon 🚀 and definitely typed 🚀 )

[karak]

Is that documentation guarding?
Actually, I had already had a type patch in my(our) local, but no works of sinon itself.

[karak]

Oh, I found this fix is incomplete!

If types is used, the return-value is a plain object that has specified properties with Spy object: see https://github.com/sinonjs/sinon/blob/master/lib/sinon/spy.js#L53

In example:

  const descriptor = sinon.spy(RGBColor.prototype, 'red', ['set', 'get']);
  const color = new RGBColor();
  color.red = 128;
  expect(descriptor.set.callCount).toBe(1);

I'd like to have official documentation.

[43081j]

All the line you linked to is doing is setting the descriptor config (as opposed to an empty, default descriptor). This is just so it sets the getter/setter correctly through the property descriptor.

It should just return a spy like any other spy call.

Your example works because sinon returns the wrapped method, which in this case is just the property descriptor its self.

[karak]

Ah, you mean a return value should have the type of the original method, property descriptor in this case, instead of just a callable, then?

EDIT: And set/get of those properties must be sinon spy as well.

[43081j]

alright so the short explanation to this is that sinon does some hacky things when you spy on something.

behind the scenes, it just attaches a bunch of sinon things to the property descriptor. the descriptor is usually {} so you don't notice anything extra, but when you use ['get', 'set'], it is {get: fn, set: fn}. Meaning technically a spy with get/set does have a get and set property (because again it is just the descriptor with sinon stuff added to it).

we can't easily represent this in strong types, though... its doable in some way i imagine but could lead to unnecessary complexity. if its wanted enough, someone should take a stab at it but otherwise i'd personally leave it out (and somehow work around it).

[karak]

Agree. The typing would be hard and complex.

Can I show my image by code bellow? I know the descriptor is just a definition, but I can't grab the final type of that execution pass.

karak@95c18e5