ThisType is invalid when using object spread on generic object

[paul-thebaud]

Bug Report

🔎 Search Terms

ThisType Object Spread Generic

🕗 Version & Regression Information

• I was unable to test this on prior versions because I was not using this feature previously

⏯ Playground Link

Playground link with relevant code

💻 Code

function createObject<D extends {}>(def: D & ThisType<D>) { /* some code */ }

const namableObject = { name: '' };
// ThisType will be wrongly typed when using object spread
const spreadObject = createObject({
  ...namableObject,
  get thisName() {
    return this.name; // <-- Property 'name' does not exists on type '{}'
  },
});
// Correctly typed otherwise
const fullObject = createObject({
  name: '',
  get thisName() {
    return this.name;
  },
});

🙁 Actual behavior

The generic type is broken when the given object value contains a spread object.

🙂 Expected behavior

The generic type should contain the spreaded object properties as those are finally included into the result object.

Comment

This may be related to #26412 and other issues like this one but I didn't found a workarround to get a ThisType correctly typed.

[RyanCavanaugh]

I'm not immediately convinced this one isn't fixable-in-principle. The non-spread case works and there isn't a circularity, so I don't see any blockers to it working. I guess the only problem is that properties from namableObject might be overridden by properties later in the object literal?

[paul-thebaud]

Hi,
Thanks for your quick answers. I agree with @RyanCavanaugh, I dont see anything blocking here.
Yes, the namableObject could be overwritten, but the type of the currently constructing ThisType should still be inferrable IMO.

[paul-thebaud]

@RyanCavanaugh, I don't know at all how Typescript works internally. I can make a PR for this if needed but I'll need some guidance.

[RyanCavanaugh]

It's very unlikely the fix here is approachable by someone not already quite familiar with the TypeScript codebase. It's definitely hard enough that I can't offer any concrete guidance on where to start.

[paul-thebaud]

OK no problem. I'll wait for any update on this and notify this issue in my package for now. Thanks a lot for those explanation and triage.