Use this checklist to validate a build on a disposable test setup before relying on it. GitHub Actions checks syntax, compilation, and packaging; it does not execute these Windows storage-behavior tests.
- Use a test PC or disposable Windows installation where possible.
- Back up important data and record current drive-letter assignments.
- Do not use a disk containing irreplaceable data for destructive or ambiguous test cases.
- Run the manager elevated. Keep the release ZIP and build artifacts separate from any existing installed copy.
- Record the Windows version, release/tag or commit, test result, and relevant
PersistentSD.logentries. - Confirm you can recover from a failed test before testing installation or drive-letter changes.
- Launch the manager as Administrator.
- Refresh the disk list; confirm device names, sizes, and IDs are understandable.
- Save a device selection and preferred letter, close the manager, reopen it, and confirm the selection persists.
- Confirm the script-location preference persists after closing/reopening.
- Confirm the UI rejects missing devices, missing letters, and unsupported letters.
- Confirm Check Status clearly reports a missing task and an unresolved script path.
- Confirm Remove Task requires confirmation and leaves script files/configuration intact as documented.
- Install with no existing task and a valid destination.
- Confirm the task exists, its action points to the generated script, and it runs as SYSTEM at startup.
- Update an existing task in the same destination; verify a backup of the old script and task XML is created.
- Update an existing task whose script is in the other supported destination; verify the prompt and backups are correct.
- Test an existing task whose action cannot be parsed or whose script is missing. Confirm the app does not silently report it as healthy.
- Test a read-only/unwritable destination and a task-registration failure. Confirm new preferences are not saved on registration failure and the manager attempts to restore the prior script or remove a newly created script.
- Where feasible, test a failure after task registration succeeds, such as a preference-save or script-path verification failure. Confirm the warning explains that the task may already exist.
- If recovery fails, capture the exact error and verify the task, script path, and backup folder manually.
- Confirm backups are readable and previous scripts are not deleted automatically.
- Remove the task and confirm its effect is limited to the task, without unexpectedly deleting scripts or preferences.
- With the preferred letter free, confirm the intended eligible data partition receives it.
- With the preferred letter owned by a different volume, confirm that volume keeps its letter and PersistentSD logs the conflict.
- Confirm system, boot, recovery, and reserved partitions are not selected for reassignment.
- Test the selected disk offline; confirm PersistentSD does not bring it online automatically.
- Disconnect and reconnect the selected removable disk; confirm the arrival watcher and periodic retry behave as expected.
- Start with the selected disk absent, then connect it later.
- Test a disk with multiple eligible partitions and verify the selected partition follows the documented preference order.
- Test a missing, duplicate, or changed disk UniqueId; confirm no drive-letter changes occur unless exactly one disk matches.
- Inspect logs for useful error messages when storage queries or assignment attempts fail.
- If practical, simulate watcher-registration failure and confirm the task stays running and falls back to periodic retries rather than exiting.
- Confirm the build workflow succeeds on the exact commit intended for release.
- Download the asset from the published release page, not just the Actions artifact.
- Extract the release ZIP on a clean test account or test machine.
- Confirm the ZIP contains
PersistentSD.exe,PersistentSD.ps1,README.md, andLICENSE, with the EXE and script together. - Confirm the README in the ZIP describes the same version/download flow and safety limitations as the release notes.
- Test the exact ZIP intended for publication, not only an EXE built locally.
- Use a new tag/release for changed binaries; do not replace an existing version's asset.
- Release notes must distinguish successful compilation from actual Windows runtime testing.
For each failed or skipped item, record the step, Windows version, exact build/release, expected result, actual result, and relevant log output. Mark tests as passed, failed, or not run. Do not describe runtime behavior as verified unless that exact behavior was tested.