unmount regression in 3.4.0

[ZebraFlesh]

Current behavior

Using enzyme 3.3.0, it is possible to do the following:

const wrapper = mount(<MyComponent />);
expect(wrapper.find('.my-component')).to.exist;
wrapper.unmount();
expect(wrapper.find('.my-component')).to.not.exist;

When upgrading to 3.4.0, the last line now fails with Error: Method “name” is only meant to be run on a single node. 0 found instead. Indeed, adding console.log(wrapper.name()) after the unmount call results in the same error.

Expected behavior

Unmounted wrapper continues to function as expected in 3.3.0

Your environment

API

• shallow
• mount
• render

Version

library	version
enzyme	3.4.0 (also applies to 3.7.0)
react	16.4.1
react-dom	16.4.1
react-test-renderer	16.2.0
adapter (below)	1.6.0

Adapter

• enzyme-adapter-react-16
• enzyme-adapter-react-16.3
• enzyme-adapter-react-16.2
• enzyme-adapter-react-16.1
• enzyme-adapter-react-15
• enzyme-adapter-react-15.4
• enzyme-adapter-react-14
• enzyme-adapter-react-13
• enzyme-adapter-react-helper
• others ( )

[ljharb]

I'm a little confused; what calls .name()? It might be your to.exist matcher; what happens if you use something more explicit?

[ZebraFlesh]

I'm using chai-enzyme for that. Digging into that code, it seems like that may be the source of the issue: chai-enzyme is checking that the enzyme wrapper.length > 0, failing, and then trying to report it (likely using the name). However, this a change that I noticed during debugging:

• v3.3.0: mounted enzyme wrapper has length 1 after unmount
• v3.4.0: mounted enzyme wrapper has length 0 after unmount

So is that an enzyme regression, an intended change, and/or an implementation detail which chai-enzyme should perhaps not be leveraging?

[ljharb]

It looks like the change was made in #1498 - specifically, this line was removed, which (imo incorrectly) forced the length to be 1.

In other words, I think it's an intentional bug fix - and as far as chai-enzyme goes, i think they were unintentionally relying on the erroneous behavior to have .name() work.

Could you rewrite your test case to use .name() and not chai-enzyme, with what happens in 3.3 vs 3.4? We may still be able to fix that particular method, which would unbreak your chai-enzyme use case.

[ljharb]

@ZebraFlesh closing; happy to reopen if you can make the change in my previous comment and there's something we can fix.