tobiasdeiminger added a comment.
Sorry for the regression! I don't have a development environment for okular
atm., the following is only from top of my head...
> Anyone knows why he would have thought of requiring a handle for showing
the tooltip?
MouseAnnotation can be in a state where an annotation has been selected by
left-click, but then the mouse went further somewhere else on the document -
maybe even over another yet unselected annotation. We do not want to show a
tooltip for the selected annotation (m_focusedAnnotation), but for that one
where the cursor is actually over (m_mouseOverAnnotation). To distinct those
cases, MouseAnnotation can internally look at m_handle. m_handle tracks the
current position of the mouse while it represents the different functional
areas of an annotation under the cursor (upper left corner, upper left corner,
inside, nowhere...).
bool MouseAnnotation::isMouseOver() const
{
return ( m_mouseOverAnnotation.isValid() || m_handle != RH_None );
}
IIRC, isValid() and m_handle != RH_None are not exactly equal. The latter
condition is true if the cursor is inside the imaginary surrounding rectangle,
while isValid() is only true if mouse is over actual graphics. That makes a
difference when you have an annotation like a circle, where only the border is
drawn, and the inner area is not part of the annotation.
It seems good to remove the lines as you're doing it with your patch, as they
seem to be redundant. We do the same check right before in pageView.cpp by
calling MouseAnnotation::isMouseOver():
if ( d->mouseAnnotation->isMouseOver() )
{
d->mouseAnnotation->routeTooltipEvent( he );
}
Does removing the lines already fix the bug? I'm not sure if it can, it
removes redundancy and makes the code cleaner - but does it also change
effective behavior?
REPOSITORY
R223 Okular
REVISION DETAIL
https://phabricator.kde.org/D7874
To: aacid
Cc: tobiasdeiminger, #okular, aacid
