[Types] useMutation returning union undefined with new types?

[Buuntu]

Describe the bug
Mutate now potentially returns an undefined type, which is different from the return type before I believe? Maybe that's the expected behavior but by looking at the docs you generally aren't checking that something is returned. Shouldn't it just throw an error rather than return undefined if something goes wrong?

So using the new types you now have to check that the data exists before using it.

Before

const [mutate] = useMutation(mutateFunction);

try {
  const data = await mutate();
  // do something
} catch (err) {
// error handling
}

After:

const [mutate] = useMutation(mutateFunction);

try {
  const data = await mutate();
  if (data) {
    // do something
  }
} catch (err) {
// error handling
}

There might be a good reason for this but it does deviate from the docs (https://react-query.tanstack.com/docs/guides/mutations#basic-mutations) and previous types.

[MichaelDeBoey]

Is this still a valid problem now #767 is merged? 🤔

[Buuntu]

Is this still a valid problem now #767 is merged? 🤔

I just installed it today so I think so? WIth:

npm i react-query@tsnext

My package.json:

"react-query": "^2.5.7-tsnext.1",

[boschni]

Hi @Buuntu, thanks for giving feedback on the new types! The type has indeed changed because the mutation returns undefined when an error occurs and the throwOnError option was not set to true. The documentation does seem to imply that the mutate function will throw by default because of the try catch in the example. This could either be a bug in the code or in the docs. What should be the default behaviour here @tannerlinsley ?

[tannerlinsley]

Mutate does not throw by default unless throwOnError is true. The docs should be updated to reflect this.

[Buuntu]

I suppose it's not doable to set a conditional type depending on what the config is? Something like: https://www.typescriptlang.org/docs/handbook/advanced-types.html#conditional-types.

If not I'm happy to update the docs and submit a PR.

[quantizor]

Is this still a valid problem now #767 is merged?

Yes it is, I get the error on 2.5.11 but not on 2.5.7-tsnext.1

[tannerlinsley]

Looks like with the docs and the new 2.5.7-tsnext.1 types, we'll be in good shape.