[marcelgerber]

Turns out we never adapted our tests for the altered behaviour introduced in #11869.
I tried my best to preserve the test cases.

cc @abose

[petetnt]

All tests pass, changes look fine, LGTM

[ficristo]

It seems to me that the order of tests will matter now, so if we will start to shuffle around the test code we could start to see some failures.
I wonder if we could do something more explicit about it...

[redmunds]

@ficristo You're referring to the "remember the last search behavior", correct? At the very least, each test should store previous value (in beforeEach), then restore it after the test (in afterEach). A better approach would be to mock the mechanism that stores the last search, and then create/destroy a new mock for each test.

[ficristo]

@redmunds yes.
Maybe for now we can commit this and file a new issue about this thing: what do you all think?

[abose]

Oops, Missed this test failure. Thanks @marcelgerber
The order of the tests will come into picture when the test cases are modified, in which case the author who writes the test will mostly become aware of this quirk. As this will put the test suite back to working state, Merging the fix for now.