Verify a desktop installation

For contributors and package maintainers: use this checklist to verify a release on a real desktop. It covers login, input and visible behaviour that headless tests cannot prove. Ordinary installation ends with the steps in the install guide.

On this page

1. Log in
2. Check your shell
3. Check windows and keys
4. Reload
5. Lock and unlock
6. Screen sharing and remote input
7. Return to GNOME

1. Log in

Use the package guide or complete source session registration. Choose Gnoblin at the login screen.

Record the build you installed, distribution, GPU, displays and scale factors.

2. Check your shell

Verify the bar, dock, launcher and notifications appear and respond to input. Open and close menus near each screen edge. Test every display.

If the shell fails, confirm desktop right-click opens a terminal and the recovery panel appears. Check the shell's log before restarting it.

3. Check windows and keys

Test focus, typing, minimise/restore, maximise, fullscreen, drag and resize. Check app switching, launcher shortcuts and any custom bindings.

For SSD windows, check all buttons, titlebar drag and edge resize. A correct screenshot alone does not prove input works.

4. Reload

Change one visible config setting, then:

Shell

gnoblinctl config reload

Confirm the setting changed and existing windows survived.

5. Lock and unlock

Confirm the lock screen receives input, desktop controls are unavailable while locked, and the shell recovers after unlock.

With the desktop portal backend installed:

Start a share and complete the backend's consent dialog.
Confirm the selected monitor is the one actually shared.
Test remote input and confirm the backend's device prompts.
Stop sharing and confirm the stream ends.

Gnoblin only routes the portal request. The selected backend decides access.

7. Return to GNOME

Log out and select GNOME. Confirm its desktop still works. For package-release testing, also remove Gnoblin using its documented uninstall path and check that the Gnoblin login entry disappears.

Record failures separately from skipped checks. Include the exact commands, logs and visible outcome; a successful build is not a successful desktop test.