I've been using Enzyme for a couple of years, and I've been happy with most of the time. But it happened to me a couple of times that I misread a snapshot, and a bug ended up in production!
Also, I've been teaching React to some colleagues lately, and I've realized it is not easy to understand Enzyme.
Add to this facts that I like to learn new stuff and use it in project always, and you'll understand why I switched recently to react-testing-library.
It is a different way of doing tests, tbh. People tend to understand it earlier because it is more similar to other unit tests techniques.
The only tradeoff (if you can call that) is that I have to mock way more stuff using react-testing-library than using Enzyme. It makes tests more verbose, but... I think I like it better! It helps me isolate what to test from a component.
So, that's my experience so far. Hope you find it useful.
From my experience, you'll have to rethink how to test stuff most of the time. A direct conversion from Enzyme to RTL won't work.
I always try to avoid testing implementation details. I don't have much details about your example, but in RTL is quite easy to get/set fields content, then you have to find the submit button, fire the click event, and expect some data to be sent.
I found way more difficult to test render props for instance...
And its hard to find material to study, I cant understand the Kodds examples too much :/
And I hate the fact that I cant find a comp by id or name hehe
If you are using 3rd party components for inputs/labels, that's perfect! You don't need to test that whatever they do internally works, you just need to test that your component does what it needs, so you can mock them.
jest.mock('ui-library/Input', () =>
jest.fn(({value, handleChange}) => <input data-testid="input" onChange={handleChange} value={value} />)
);
it('should do magic', () => {
const {getByTestId} = render(<Component />);
// we test that the component is rendering
expect(getByTestId('input').value).toBeInTheDocument();
// we test that the component shows the proper value
expect(getByTestId('input').value).toEqual(oldValue);
// we test that Component can handle change events
// (we don't care how Input does that internally)
fireEvent.change(getByTestId('input'), newValue);
expect(getByTestId('input').value).toEqual(newValue);
});
The code might have errors, but the I think the idea is clear enough!
The more u mock the less confidence u will have that all is well in the app/component.
I'd favor more integration type tests (that do less mocking , perhaps only mock out network/rest api cals) over 'unit' tests. As very unit versus integration is a thorny subject :)
I used to use Enzyme for a long time in my projects. However, now it is on deprecation path and will not be compatible with future versions of React. Therefore, I decided to write a tool that will help me and my team to migrate our projects to RTL. Check my article: dev.to/srshifu/migrate-away-from-e...
It has links to the tool you can download and use in your project
Top comments (15)
I've been using Enzyme for a couple of years, and I've been happy with most of the time. But it happened to me a couple of times that I misread a snapshot, and a bug ended up in production!
Also, I've been teaching React to some colleagues lately, and I've realized it is not easy to understand Enzyme.
Add to this facts that I like to learn new stuff and use it in project always, and you'll understand why I switched recently to react-testing-library.
It is a different way of doing tests, tbh. People tend to understand it earlier because it is more similar to other unit tests techniques.
The only tradeoff (if you can call that) is that I have to mock way more stuff using react-testing-library than using Enzyme. It makes tests more verbose, but... I think I like it better! It helps me isolate what to test from a component.
So, that's my experience so far. Hope you find it useful.
I'm learning RTL and I use enzyme. I Try to reproduce the same tests that I use on enzyme to RTL, but I cant.
For example, how could I check if all form fields was filled and compare with a object that I'll send to backend?
From my experience, you'll have to rethink how to test stuff most of the time. A direct conversion from Enzyme to RTL won't work.
I always try to avoid testing implementation details. I don't have much details about your example, but in RTL is quite easy to get/set fields content, then you have to find the submit button, fire the click event, and expect some data to be sent.
I found way more difficult to test render props for instance...
hmmm I see.
And its hard to find material to study, I cant understand the Kodds examples too much :/
And I hate the fact that I cant find a comp by
id
orname
heheYou'll find the way, no worries...
I ended up using tons of data-testid in the mocks (I try really hard to avoid them in the real code).
I wanna try to avoid using it too, but how I use 3rd party components, I can't use
htmlFor
for example, and I rarely use placeholderIf you are using 3rd party components for inputs/labels, that's perfect! You don't need to test that whatever they do internally works, you just need to test that your component does what it needs, so you can mock them.
The code might have errors, but the I think the idea is clear enough!
Yeah, its clear, thank you bro
"And I hate the fact that I cant find a comp by id or name hehe"
I don't know if you already figured out how to do it, but I actually can by doing, for example:
const { container } = render(Component);
expect(container.querySelector("#myid")).not.toBeNull();
You can create your custom selectors like this:
I have some custom selectors created in a utils file (name, type, id...).
The more u mock the less confidence u will have that all is well in the app/component.
I'd favor more integration type tests (that do less mocking , perhaps only mock out network/rest api cals) over 'unit' tests. As very unit versus integration is a thorny subject :)
have u any example ?
Favour Integration Tests - great article by the famous Kent C Dodds. His article says it better than me :)
I think that article moved here: Write tests. Not too many. Mostly integration.
I used to use Enzyme for a long time in my projects. However, now it is on deprecation path and will not be compatible with future versions of React. Therefore, I decided to write a tool that will help me and my team to migrate our projects to RTL. Check my article: dev.to/srshifu/migrate-away-from-e...
It has links to the tool you can download and use in your project