cases when not unmounting causes render failures in storybook and storyshots

[jrmurad]

In my decorator, I reset global state to empty with redux.

In each story, I dispatch actions which set the required global data for the component to be rendered.

When I use my browser's refresh while on such a story or go directly to its URL, I get an error page because it:

1. runs story fine
2. for some reason, runs the story again
3. this causes the decorator's reset to run
4. the reset clears required data from the component so it breaks

I believe the correct thing to do would be to either (a) not run the story twice or (b) unmount first before running again.

Two different hacks got this working for me.

1. Global decorator with:

    if (typeof document !== 'undefined') {
        ReactDOM.unmountComponentAtNode(document.getElementById('root'));
    }

2. Removing the guard put in for Action logger breaks stateful components #116

A related issue occurs when using storyshots. The previous story component does not unmount between stories. I was able to fix this by modifying test-bodies.js to unmount. I can submit a PR if this is an appropriate solution.

export const snapshotWithOptions = options => ({ story, context }) => {
  const storyElement = story.render(context);
  const testInstance = renderer.create(storyElement, options);
  const tree = testInstance.toJSON();
  expect(tree).toMatchSnapshot();
  testInstance.unmount();
}

For reference, this is how my stories look:

import {dispatch} from 'store';
import {reset} from 'ducks';

import SomeComponent from 'containers/SomeComponent'

storiesOf('story', module)
    .addDecorator(story => {
        dispatch(reset());
        return story();
    })
    .add('pending removal', () => {
        dispatch(...); // if data set here is required by SomeComponent, we have problems
        return <SomeComponent />;

[stale]

Hi everyone! Seems like there hasn't been much going on in this issue lately. If there are still questions, comments, or bugs, please feel free to continue the discussion. We do try to do some housekeeping every once in a while so inactive issues will get closed after 90 days. Thanks!

[jrmurad]

Note: this is still occurring with 3.3alpha and the same workaround handles it for me.