camera/multiple-resolution-images_* job is unable to run on the server and core.

[rickwu666666]

Bug Description

camera/multiple-resolution-images_* job is unable to run on the server and core. This is because it is located in the client-cert-iot-ubuntucore-20.pxu / camera-automated and may require refactoring of the logic in camera_test.py.

Based on this part of the code, if resolutions is provided, Clutter and Gtk will be initialized. However, both of them require a GUI for initialization. And that causing the job to fail.

 if args.test in ['display', 'still', 'led', 'resolutions']:
    import gi
    gi.require_version('Gst', '1.0')
    from gi.repository import Gst
    gi.require_version('GLib', '2.0')
    from gi.repository import GLib
    gi.require_version('Clutter', '1.0')
    from gi.repository import Clutter
    gi.require_version('Gtk', '3.0')
    from gi.repository import Gtk
    Gst.init(None)
    Clutter.init()
    Gtk.init([])

However, we tried to work around the issue by running the function without initializing Clutter and Gtk, and it seems to work fine.
https://certification.canonical.com/hardware/202305-31566/submission/323229/test/200875/result/34817873/

To Reproduce

1. A machine support camera with a camera attached.
2. Run job camera/multiple-resolution-images.* on server or core
3. Job will fail while initial clutter

Environment

• OS: Ubuntu 22.04.2 LTS
• Checkbox Type: Snap
• Checkbox Version: checkbox22 2.8 rev 437
• Hardware being tested: devices

Relevant log output

 (camera_test.py:15926): Clutter-CRITICAL **: 07:01:18.542: Unable to initialize Clutter: Could not initialize Gdk
 (camera_test.py:15926): Gtk-WARNING **: 07:01:18.543: cannot open display:

https://certification.canonical.com/hardware/202305-31566/submission/323043/test/76060/result/34787566/

Additional context

No response

[pieqq]

The things to take into account are:

• the script should work with desktop images (with a GUI) and with server/core images (not Gtk available)
• MIPI camera don't seem to work with fswebcam (see Fix: MIPI camera can use gstreamer when fswebcam can't take photo #264 )

It should be possible to use gstreamer without Gtk, that is to perform the image capture (and save it to file) without using Gtk, but only Gst.

[pieqq]

camera_test.py comes with the following commands. Some of them are used in a graphical environment (desktop) and require some interaction with the operator, some don't, so we can split them like this:

Command	Desktop	Server/Core
detect	Required	Required
led	Required	Not required
display	Required	Not required
still	Required	Not required
resolutions	Required	Required

To-do:

1. Investigate what parts (detect,led,display,still,resolutions) really require Gtk/Clutter to run
2. Isolate the use of Gtk/Clutter to these
3. Add some tests
4. Test the changes on 22.04 classic, 22.04 server and UC22 (using side-loaded provider should be enough)

If done well, it might be possible to remove fswebcam from the script entirely.

[syncronize-issues-to-jira]

Thank you for reporting us your feedback!

The internal ticket has been created: https://warthogs.atlassian.net/browse/CHECKBOX-1200.

This message was autogenerated

[zongminl]

Fixed by #1400