It is considered bad practice, and mainly a lazy way of disconnecting
signal handlers without tracking individual handler IDs.
We can do better by using connectObject(), which provides the same
level of convenience without the dodginess of getting behind the
garbage collector's back.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/275>
Provide all licenses used in the project in a LICENSES folder and
add SPDX license and copyright information for all files in
accordance with the Reuse Software[0] specification.
The copyright information is based on the file's git history,
using a fairly generous definition of "non-trivial".
As of the spec recommendation, the information is generally added
as comments in the files themselves, except for
- NEWS, README and similar top-level standard files, so that
a SPDX code isn't the first thing people encounter
- files that don't support comments (json) or where they'd
be a bit awkward (.desktop, .service)
- anything under po/, to not interfere with translation teams
Those are covered by a .reuse/dep5 files, except for image assets,
where separate .license files are used (It would be possible to
add comments to SVG files, but I don't trust image editors to
preserve them).
[0] https://reuse.software/
Part-of:
<https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/224>
As gnome-shell is moving to ESM, it will now load extensions as
standard modules instead of using legacy imports. The change boils
down to exporting the Extension class as default, but we can also
start using standard imports for introspected modules now, so do
that at the same time.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/259>
Now that classic styling is based on color scheme instead of
a dedicated "classic" stylesheet, we should do the same for
extension styling, with the bonus that it also works with the
regular appearance setting outside the classic session.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/254>
gnome-shell started transitioning to gjs' object spacing rule,
i.e. `{foo: 42}` instead of `{ foo: 42 }`.
We have a much smaller code base than the shell and aren't using
a secondary "allowed-but-deprecated" configuration that allows a
gradual transition, so just pull the switch and update to the new
style.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/240>
Currently, the visibility of the window list on a given monitor is set
according to the fullscreen status of the primary monitor. When a
fullscreen application is on a secondary monitor (but not on the primary
monitor), entering and exiting the overview will lead to the window list
incorrectly being visible on the monitor with the fullscreen
application.
Instead, determine the visibility based on the fullscreen status of the
monitor being evaluated.
Fixes https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/issues/400
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/230>
As a side-effect of supporting class fields, regular constructors
now work in GObject subclasses. Using _init() still works and
there's no functional difference, but it's simply much nicer
to use the same syntax for all classes.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/215>
The reveal animation moved from Main.layoutManager.keyboardBox to
the keyboard itself, so instead of applying an additional translation
for the bottom panel, we override the translation that would reveal
the keyboard (and thus prevent it from showing altogether).
Fix this by moving our translation to the keyboardBox instead.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/199>
Mutter uses an undefined initial in-fullscreen state, so it will
always emit the `in-fullscreen-changed` signal when it determines
the actual initial state.
This didn't use to be an issue when the shell started in the session,
but now results in the window list ending up visible in the overview
on startup.
Work around this by hiding ourselves again when the in-fullscreen
state changes in the overview.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/185>
The helper functions date back to a time when AppButton and WindowButton
were unconnected classes. But nowadays they share a common base class, so
we have a better place for them than external helper functions.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/180>
Since commit a6ee142f21, the extension archives that are uploaded
to extensions.gnome.org only contain strings that are relevant for
the extension, not all translations from all extensions.
Unfortunately all extensions still share a common gettext domain,
so the extension with the last bind_textdomain() call wins and
leaves the others without translations.
We'll address this by using distinct domains when not installed
system-wide. That becomes easier if there is a canonical place
for the text domain, with the existing metadata key being the
natural choice.
https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/issues/335
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/179>
The panel-button introduces some horizontal padding which is insensitive
to scroll events. Without this change, there is a small dead zone in the
corner that cannot be used to switch workspaces with the mouse wheel.
For useMenu mode, this has the effect of removing all of the horizontal
space to the edge of the screen, so I add some back with the
status-label-bin margin.
This a is similar change to 8bad8a3b63.
Fixes#315.
Part-of: <https://gitlab.gnome.org/GNOME/gnome-shell-extensions/-/merge_requests/171>
