> > > For example, QML has "anchors", and you *can* do layout very easily, and > > even parent-item-resizing-based-on-child. However, layout is > fundamentally > > "wrong" when you start moving into "states" and "transitions". Layout > > fundamentally "fights" your ability to do rich presentation for today's > > sophisticated users limited only by precious screen real-estate. >
Ville respondeth: > Not necessarily - e.g. you can use e.g. AnchorChanges and > AnchorAnimation to keep the rich possibilities and still deal with > layouts. > Good point. I'm asserting, and struggling with, with the logical concept that even though QML is (mostly) real-estate "thingies" <insert disclaimers and exceptions>, its paradigm does not imply a One:to:One for ParentContainment:to:DivisionOfRealEstateAmongChildren. It *can* do that, but unlike widgets, it doesn't *force* you to do that. That's what layouts do, and that's what anchors do, and that is often what we want (declarative subdivision of real estate). (I happen to like the anchors more than the layouts, since the anchors are far more flexible. Having said that, I'm still writing my own layout class(es)). However, the "GraphScene" presentation concept similarly implies containment of children within parents (each QGraphicsItem can have a hierarchy of parents and children), but that is *logical* containment that (can) have *nothing* to do with subdivision of the parent's real estate. It's possible, and useful, for the item's children's bounding box to be (mostly) irrelevant as a display concept. In that case, items become "independent actors", including control of their own placement and mobility, including responses to input, or reflection of state from the back-end model. Kind of like teenagers that no longer listen to their parents, but whom still love their parents, and may possibly show up at the parent's house for the Holidays or occasionally for a hot meal. The "top-down" paradigm of real estate subdivision and control, even if flexible, seems to require different thinking about the design and presentation metaphor than that from a world of "independent actors". The reason I'm scratching my head with this is that for my current work, I'm attempting to go "bottoms-up" with (almost only) independent actors -- a lot of them -- with very little policy regarding real estate -- since they reflect tremendous implied volatility. In fact, *because* there is tremendous implied volatility, QML seems so *ideal* (I don't need to pre-ordain everything, or even *imply* anything, even when I pre-ordain some parent/child relationships). (e.g., Example is in reference to animated presentation of a build system, and test regression management system.) I forgot about AnchorAnimations, though. Nice. --charley
_______________________________________________ Qt-qml mailing list Qt-qml@trolltech.com http://lists.trolltech.com/mailman/listinfo/qt-qml