--fix showing advanced settings window in order to update
the window geometry ONLY WHEN there are availableScreenGeometry
changes from Latte::View(s).
--As a side effect changing docks/panels location under wayland
when they have enabled their background blur has reduced crashes
during location changes.
--improve the new worksarounds in order to always
apply the activities that a view and its windows are
present when the workarounds were triggered from
kwin faulty behavior. The kwin faulty behavior is
that when any activity is paused then all Latte::Views
become hidden and lose their activities references.
The workarounds are timers that when they trace
a faulty visible state !visible are restoring the
normal behavior to views.
--trying an alternative approach concerning real floating
Views and screen edge triggering. In such case the user
has triggered to show a REAL FLOATING VIEW, it is considered
normal from user point of view for the view to remain shown
until there is windows changed state or the mouse has
escaped the View main window area
--indicators are now tracked only based on their
installation path, that means that are updated
more precise when changes are applied and
at the same time views are recreated only when
the show the specific indicator that was changed
--improved settings for Floating windows in Behavior
tab and add a new option to hide ALL screen gaps meaning
both length screen and thickness screen gap when there
is a maximized window in the screen
BUG:415630
--following the new design applied first at
View:availableScreenRegionChangedFrom instead of different
View parts to trigger directly m_corona relevant signal,
all View parts are triggering the View relevant signal
and as such the can be disabled all together
easily.
--when as Always Visible view was unloaded from
a CentralLayout because its Activity was closing
it was creating a crash. This patch fixes the
situation by releasing sensitive view signals
earlier.
--when the application launcher is not a popup then
we do not have to wait for the dock show we can execute
the activation code immediately
--when Latte activates an entry can now understand
if a popup is going to be shown in order to delay
the execution. If a popup is NOT going to show either
for a latte task or an applet then the code is executed
immediately
BUG:415417
FIXED-IN:0.9.6
--all the qml handling code from globalshortcuts is
now moved to Latte::View::ContaimentInterface. This
way the code is cleaner and can be expanded easier.
--draggingStarted signals/slots were used in order
to handle the active window dragging from empty areas.
The new approach ungrabs the mouse properly through
View relevant function and at the same time sends a
QEvent::Leave event to inform applets that the mouse
has left the view. This is enough in order to remove
any older draggingStarted workarounds and hacks.
--when a vertical view is in busy state and is touching
also a horizontal view that is fully transparent in
SMART mode then the visual result is not that good.
In that case it is better to show the view background.
The horizontal view background in that case can provide
three different states:
a) if the panel style for popups is activated then the
horizontal view will follow the plasma original style
b) if NOT [a] and solid background for touching windows
is enabled then solid background will be offered
c) if NOT [a] and NOT [b] then the isBusy state
transparency will be offered
--trying to increase smartness for horizontal
and vertical views. In multiple layouts environment
a view can exist in cases that one or more of
the sided views that is touching it is NOT
transparent, it would be nice for those
cases the top or bottom view to LOSE also
its transparency in order to not look
inconsistent.
--when KWin edges are used in a SharedLayout then the
all view Layout::isCurrent() function must be used
otherwise the layout is not identified properly as
current based on different layouts per activities
--in such case under x11, the winIds of the visible
Latte windows are changed and as such the windows
tracking mechanism becomes broken. The code introduced
tries to track the case the update the WM::ignoredWindows
properly
--this function creates irrelevant crashes even
under x11 environment... Dropped in favour of
show()/hide() functions that seem to not create
any issues
--fixes to signals in ViewParts in order to avoid
crashes. Currently the recreateView approach is used
in order to reload indicators runtime dynamically.
The "View::setSource" approach does not reload the
indicators properly
--View::WindowsTracker now provides all the time
information about CurrentScreen and AllScreens!
This way all consumers (Latte::Containment, Applets etc.)
can choose whatever suits them best.
--when dragging a new applet over the view
the responsiveness is now improved vastly
--the new architecture uses two different
DragAndDropAreas. One which is the old one
and is used when dragged launchers, urls and
separators. The second is used only when
applets are dragged into the view
--the visuals for adding launchers/applets have
been improved and a circular visual is preferred
now instead
--fix all events identifications from DragMove and
DragExit etc. in containment::qml part and all drags
flags are calculated only at DragEnter! This fixes
the DragMove signaling and all DragMove signals are
sent correctly.
It needs to be fixed also for Tasks plasmoid because
every time the mouse hovers the Tasks Plasmoid then
the DragMove signals are NOT SENT as they should.
--any view and view::containment() signals that
are related to layout are now moved in the
View::setLayout() in order to be reinitialized
properly when the view changes to another
layout
--this is side-way path for mouse events and
more specific the mouse pressed, released
events to be sent from parent window to its
qml containment part. This way it is possible
to keep the original plasma applet behavior
such as original plasma tooltips and provide
Latte specific functionality such as animated
clicks from indicators.
NOTE: An investigation - testing took place in
order to check if this approach could be also used
for mouseMove events to be used for applets parabolic
effect. The result was that this was not possible because
there were too many signals and too much calculations
too identify for which applet the signal belonged to.
The parabolic effect was not fluid in that case and
could not catch up.
Michail Vourlakos