Thanks for the fix. I'm gonna try to figure out exactly why it was failing before merging in order to lessen the risk of unintended side effects. I also want to add jobs in CI that test against jest 25 & 26. Should hopefully have some time to do so today.

I've added jest@26 to the build pipeline and the test spass on master. Are you able to set up a reduced test case illustrating the situation when things are breaking for you. Pasting in a comment here is fine

Here is the gist of the test case that is failing:

    it('should handle address input changing', () => {
      const handleAddressChange = jest.fn();
      const { getByLabelText } = render(
        <AddressAutocomplete
          name={name}
          label={label}
          address={''}
          handleAddressChange={handleAddressChange}
          {...i18nProps}
        />
      );
      const input = getByLabelText(label, { selector: 'input' });
      const value = '123';
      const url = `${TEST_URL}/actions/addressPredictions?filter[address]=${value}`;

      const addressMock = fetchMock.mock(
        url,
        { status: 200, body: [] },
        {
          method: 'GET',
        }
      );

      fireEvent.focus(input);
      fireEvent.change(input, { target: { value } });
      jest.runAllTimers();

      expect(addressMock).toHaveBeenCalled();
      expect(handleAddressChange).toHaveBeenCalledTimes(1);
      expect(handleAddressChange).toHaveBeenCalledWith({ address: value });
    });

The actual component implementation just debounces the search value and then executes a redux dispatch which in turn triggers the fetch. Not really much special going on. Unless it's something weird with using the timers and delaying the fetch call.

@wheresrhys - could it potentially be a node version issue? I'm running v14.

Is the test running in the browser? Or in a browser-like environment (e.g. JSDom) and mocking the fetch global? .sandbox() isn't meant to be used in that case.

I'll probably need to see the failing test + all the test runner set up in order to properly debug

These tests are running as part of a create-react-app, latest version of react-scripts and its jest configuration, along with jest 26, so using JSDom.

I only mentioned the sandbox because I thought it was relevant to some failing tests I had locally, but I now see that all of the test cases use the sandbox method.

There are a couple other cases in our codebase that do not do any debouncing or timer related business, and those also have the same issue, so I don't think it's anything there like I suggested earlier. These are simpler functions that just call the browser fetch, similar to most of the test cases you already have.

Edit: I'm not using sandbox anywhere in my code, just doing import fetchMock from 'fetch-mock-jest'; in the test and using the browser fetch like normal in app code.

I don't really have at the moment to set up a full environment to replicate. If you could create a minimal test case repo I could clone (or are able to share an existing repo with me) that'd be a great help

Sorry for the delay in getting back with you, I'm working through a separate upgrade issue with another library since I found a workaround for this. I should hopefully have that wrapped up today / tomorrow, then I plan on putting together a codesandbox to try and replicate the issue.

@wheresrhys @green-arrow don't forget about this one! Based on your conversation, it sounds like y'all are close! 🤞