--The new implementation for snapping needed to remove
the touching view edge from touching algorithm. The
windows trakers now provide two separate values for each
touching.
----First value is for regular window touches for which
their window geometry intersects with view published
absoluteGeometry
----Second value is for windows that are touching
view edge without intersecting with view geometry
--when a behaveAsPlasmaPanel is sliding-in and View
is in editMode, it is a good idea to calle
Positioner::syncGeometry() in order to validate
the View window geometry in edit mode. The problematic
case example was a right SideBar Panel that when
was hidden and the user was using Meta+A to
show its settings, the View did not change its
geometry properly in the end.
--View::Effects now provides in API in order for
elements to publish regions that need to be added
or removed from View::mask(). A good example for this
is View Config windows that must remove a rectangle
for their bottom edge in order for users to have
fully access to their action buttons at the bottom.
--when hidden configuration windows are deleted memory
consumption is improved and provides much lower values
--when hidden configuration windows are not deleted
memory consumption increases but their responsiveness
for hiding/showing is instantly
--really handy and beautiful in order to confirm the
calculations that are valid. Already a bug was found
for ParabolicEffect. There is 1px difference between
hovering Tasks and Applets
--View recreation path was broken from new implementation
for Latte::Interfaces that pass objects to containment
interface. This is now fixed and Interfaces::View object is updated
and broadcasted correctly when changed
--when GtkFrameExtents are zeroed for behaveAsPlasmaPanels
then it is better to recreate the view to avoid freezes and
hidings of the view because of compositor strange behavior
--use a single slot when View triggers its workarounds
for the kwin issue that kwin hides panels when an activity
stops. In such case the view shadows are also redrawn.
--there are cases that the Blurred area svg returns
a NULL mask() even though it should not. In such
cases the workaround takes place and for the workaround
the calculations have already taken place and we
do not need to reapply them because in that case
the blurred area moves out of the window.
Michail Vourlakos