Bug#960696: chrome-gnome-shell: /usr/share/applications/org.gnome.ChromeGnomeShell.desktop is misconfigured

[Simon McVittie]

Control: reassign -1 chrome-gnome-shell,src:kservice
Control: found -1 kservice/5.62.0-1

On Fri, 15 May 2020 at 22:08:35 +0800, Jason L. Quinn wrote:
> Start of any QT application from the console produces the following chrome-
> gnome-shell related error message:
> 
> ====================
> kf5.kservice.services: The desktop entry file
> "/usr/share/applications/org.gnome.ChromeGnomeShell.desktop" has Type=
> "Application" but no Exec line
> kf5.kservice.sycoca: Invalid Service :
> "/usr/share/applications/org.gnome.ChromeGnomeShell.desktop"
> ====================

>From codesearch, this seems to be coming from the kservice source
package. It isn't clear to me whether this is a chrome-gnome-shell
bug, a kservice bug, or a bit of both, so I'm assigning it to both
packages for now.

> The probable fix is that "Service" is probably intended instead of
> "Application" for the Type.

That type is not documented in the Desktop Entry Specification
<https://specifications.freedesktop.org/desktop-entry-spec/desktop-entry-spec-latest.html>
although there is a note that it exists as a historical KDE-specific
extension.

It looks as though the intention is that this application is only runnable
via D-Bus activation, and not via the traditional "run a child process"
route. The Desktop Entry Specification allows this, although does not
recommend it:

     The Exec key is required if DBusActivatable is not set to
     true. Even if DBusActivatable is true, Exec should be specified
     for compatibility with implementations that do not understand
     DBusActivatable.

(note "should", not "must")

Exec=/bin/false would presumably silence the warnings from kservice -
but chrome-gnome-shell's .desktop file seems to be valid according to
the specification, and it's here for integration with GNOME Shell (which
we know *does* implement DBusActivatable), so in practice the Exec line
will probably never be used anyway.

So perhaps this is just a kservice bug, and kservice should be more
tolerant?

Another thing that might be considered a kservice bug is that random Qt
applications probably shouldn't print diagnostic messages about unrelated
.desktop files? That seems like a recipe for spamming warnings that the
application's author cannot fix.

If these messages appear at all, I would expect them to only be printed
on request - perhaps something similar to GLib's g_debug(), which is a
no-op unless you set the G_MESSAGES_DEBUG environment variable to a
suitable value.

I also can't help wondering whether chrome-gnome-shell should be
OnlyShowIn=GNOME - I'm having a hard time seeing how a service to install
GNOME Shell extensions could possibly be useful or relevant while not
running GNOME Shell :-)

    smcv