Fix type error when queryOptions enable is set to false

[yanchesky]

Fixes #7986

[djhi]

Thanks for the error report and the PR! Would you mind trying to add a unit test for this case?

[djhi]

Thanks!

[djhi]

Thanks!

[yanchesky]

Fixing failed unit test might be cumbersome 🤔. I have run it on the local machine and out of 10 tests on unchanged code:

• 4 times all tests passed
• 3 times "should notify if passed an error with a message that makes the authProvider throw" test failed
• 2 times "should take only last change in case of a burst of changes (case of inputs being currently edited)" test failed
• 1 time both 👆 failed

[yanchesky]

I couldn't find the exact reason why the test should take only last change in case of a burst of changes (case of inputs being currently edited) sometimes fails. I've run this test only and it never failed. I suspect that it may be related to react-router-dom because right before the test fails, the error is raised An update to HistoryRouter inside a test was not wrapped in act. However, this error occurs even if the test passes so I'm not sure about this.
However, knowing that I've come up with 2 solutions that make this test suite work. (By work I mean 10 test runs without any fail)

1. Create history and pass it to the CoreAdminContext component.
2. pass disableSyncWithLocation prop to the ListController component.

If that is ok with you I will make a commit fixing it.

Snapshot of the mock requests that cause the test to fail:

"calls": Array [
    Array [
      "posts.listParams",
      Object {
        "filter": Object {},
        "order": "ASC",
        "page": 1,
        "perPage": 10,
        "sort": "id",
      },
    ],
    Array [
      "posts.listParams",
      Object {
        "filter": Object {
          "q": "hello",
        },
        "order": "ASC",
        "page": 1,
        "perPage": 10,
        "sort": "id",
      },
    ],
  ],

[ITopGun]

Hi

[WiXSL]

Could you resolve the conflict?

[slax57]

Tests might be failing because you trigger setPage(1) even though you are already on page 1, which causes store to update and might trigger a re-render.

To me, you should rather move data == null alongside with data?.length === 0 in the line below.

[yanchesky]

Thanks for the info. I will check it soon.

[yanchesky]

Changed it as you suggested and seems it's working now. However on my machine, some tests are failing, but they are failing on the master branch as well.
unit test that fail: "should apply validation to both itself and its inner inputs"
e2e tests that fail: "should validate ArrayInput", and sometimes "should have tabbable menu items"

[fzaninotto]

Thanks!