On Thu, May 31, 2012 at 6:52 AM, Sebastian Zartner <sebastianzartner@gmx.de>wrote: > > On Thu, May 31, 2012 at 11:31 AM, Robert O'Callahan > > <robert@ocallahan.org>wrote: > > > > > In all major browsers, transforms normally affect the scrollable area > > > of a scrollable container. However, during an animation or > > > transition, Webkit treats the transform as having the value it had > > > the last time it wasn't animated/transitioned: > > > http://people.mozilla.org/~roc/test_transform_scrollable_area.html > > > Presumably this is a performance optimization related to asynchronous > > > compositing. > > > > > > I think browsers should behave consistently here. Should we spec the > > > Webkit behavior? I guess that would be something like > > > "Transforms affect the scrollable overflow area as expected, unless > > > they're subject to a CSS animation or transition in which case they > > > affect the scrollable overflow area as if they had their values from > > > before they were subject to an animation/transition." > > > > > > > Actually, Webkit's behavior is more complicated than this. Event > > dispatching can (at least in some cases) cause different behavior. > > > > I've updated the testcase. Now if you hover over the animated element > > while > > it's animating, you can observe changes to the result of > > getBoundingClientRect during the animation. Curiously, changes to the > > scrollbar state are still not reflected until the end of the animation. > > > > I don't think we should spec the details of that mess. However, I don't > > know what to do instead. > > I believe what Opera and Firefox do here is what the user would expect. > Why shouldn't the scrollable area be affected by an animation? Imagine the > animated object being completely outside of the scrollable container for a > longer animation and you're not able to scroll to it. Wouldn't that be > annoying? What if you want to animate an element in from a side? Wouldn't you think it's annoying if you get scroll bars while it's not completely on the screen?Received on Thursday, 31 May 2012 18:08:27 UTC
This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:17 UTC