Test gets into runaway loop if cy.wrap is inside cy.should

[bahmutov]

Cypress v6.3.0

it('gets into a loop', () => {
  cy.wrap('bar')
    .should(() => {
      cy.wrap('foo')
    })
})

Runs away, never finishes

runaway-test.mp4

Example in https://github.com/cypress-io/cypress-test-tiny/tree/runaway

[sainthkh]

Problem solved. I'll upload the PR with why it happened tomorrow.

[sainthkh]

The problem is related with the command stack. I explained why in #15386.

[sainthkh]

I tried 2 approaches and learned that allowing cy commands inside should callback is impossible because of Cypress detected that you returned a promise in a test, but also invoked one or more cy commands inside of that promise. error. I tried to solve this problem by removing Promise inside shouldFnWithCallback, but it also failed. It might be because should command is called inside a Promise.

If you want to check that cy commands don't work inside should callback, test the code below:

it('gets into a loop', () => {
  cy.wrap('bar')
    .should(() => {
      cy.wrap('foo')
      cy.wrap('test')
    })
})

All you can see is wrap foo. wrap test is never shown. Because the error is thrown.

If you want to check the thrown error, wrap fn call with try...catch below.

cypress/packages/driver/src/cy/commands/asserting.js

Lines 11 to 24 in cb281d1

	const shouldFnWithCallback = function (subject, fn) {
	state('current')?.set('followedByShouldCallback', true)
	return Promise
	.try(() => {
	const remoteSubject = cy.getRemotejQueryInstance(subject)
	return fn.call(this, remoteSubject ? remoteSubject : subject)
	})
	.tap(() => {
	state('current')?.set('followedByShouldCallback', false)
	})
	.return(subject)
	}

---

I also thought what's the best solution for this. Showing "cy commands are not allowed inside should command" might be a good idea.

What do you think, @bahmutov?

[jennifer-shehane]

Oof, I feel like I've seen this behavior before. Ideally I would like this to just work as expected, but probably throwing an error would be better than an infinite loop. I"m not sure what the rest of the team thinks though.