* don't emit finished() multiple times
* start() is not virtual (which is why it wasn't getting called sometimes!), use updateState; also gets rid of the dirty flag hack
svn path=/trunk/KDE/kdelibs/; revision=1058157
but this is the list from what I can recall:
- removing expand animation (it is a case of grow animation)
- removing AbstractAnimation and using QAbstractAnimation as the common base
- animations are non longer factories, but real animation (so it is just no
issue to mesh then with user created animations)
- animation group will correctly revert the direction of all subanimations
(useful for rewinding an animation)
- fixed opacity effect in pulse animation (now it works correctly)
- there is not 'forward' property, but animation 'direction' (this is how
rewinding an animation must be done)
- direction now is movementDirection property (used in SlideAnimation and RotationStacked)
- the weakpointer for the actual animaiton is non longer in AnimationPrivate but
in each specialized animation class
- hide property is being removed (in progress) since it really is a case of Fade
I guess this is it, now is just a matter of iron out the remaining bugs and polish
the public API.
svn path=/trunk/KDE/kdelibs/; revision=1057993
- moved transformation reference inside of the the shadow pulse
class
- ensuring that the local pointer in pulser will correctly point to the
shadow object
- recreating the pixmap if the animated widget geometry has changed
svn path=/trunk/KDE/kdelibs/; revision=1050711
assumed that there would exist a complete copy of the animated widget to
do the pulse (where another object is kept translucent and when the animation
is triggered, it will expand and fade at same time).
But is expensive to copy the whole widget, so the idea is to draw the target
widget in a pixmap and then do the pulse effect on the pixmap itself.
It is still broken, for some reason the opacity is failing to animate (should
fix this tomorrow).
svn path=/trunk/KDE/kdelibs/; revision=1048277
one effect and the opacity property of the widget would lie.
if there will be graphics glitches it would be a qt problem that should
be resolved, not by a workaround
svn path=/trunk/KDE/kdelibs/; revision=1047550
To make possible to return to previous state, it is both necessary to use the oposite
animation factor as also the start and end value used in the animation.
svn path=/trunk/KDE/kdelibs/; revision=1043658
was triggered while the animation was running and *before*
it was over.
For while, it will just ignore the geometry calculus and return
(but ideally it should listen for the event emited when the animation
is over and then recalculate the geometries).
svn path=/trunk/KDE/kdelibs/; revision=1039630
Now the QAbstractAnimations will be cached when running an animation
(instead of creating and destroying them everytime the animation runs).
The next step is to cache *also* the configuration parameters (as aseigo suggested,
by having an event filter to detect changes in the animated widgets and reseting
the animation parameters as needed).
svn path=/trunk/KDE/kdelibs/; revision=1039531
The idea is to keep in AbstractAnimationPrivate only what is shared by all
(or at least most) of animations.
svn path=/trunk/KDE/kdelibs/; revision=1039035
