- From: Rik Cabanier <cabanier@gmail.com>
- Date: Wed, 1 Aug 2012 12:36:25 -0700
- To: "Tab Atkins Jr." <jackalmage@gmail.com>
- Cc: François REMY <fremycompany_pub@yahoo.fr>, Sebastian Zartner <sebastianzartner@gmail.com>, Julien Dorra <juliendorra@juliendorra.com>, www-style@w3.org
- Message-ID: <CAGN7qDB-fPQYZeRYar4khaMMMeS=oAi22GZrSt2mVvb=++kOiw@mail.gmail.com>
On Wed, Aug 1, 2012 at 12:23 AM, Tab Atkins Jr. <jackalmage@gmail.com>wrote: > On Wed, Aug 1, 2012 at 12:14 AM, François REMY > <fremycompany_pub@yahoo.fr> wrote: > > Oh dear, I just noticed you can’t use the 'display' property at all in > > animations; I was really thinking you could and that just the > interpolation > > process was going to be ignored. > > > > Therefore, I propose to support ‘animations’ on *any property*; the > behavior > > of an animation of an non-animatable property would be to switch the used > > value as soon as the halfway to the next keyframe is reached (i.e. the > > initial value would be mapped to -1, the final one to +1; the used value > > would be the initial one for strictly negative numbers, and the final one > > for positive numbers (0 included)). > > > > This is (strangely) not what has been done for 'visibility' but I feel > this > > is what most authors will expect for other properties; we could probably > > extend the definition used for 'visibility' to 'display' from and to > 'none' > > because it can make sense for those properties to keep the element > visible > > as long as possible. Also, if a way to animate the property is added > later > > one, the behavior will probably match more closely the change halfway > than a > > change when the next frame is reached only. > > > > This is a breaking change from the current implementations, whose some > > already shipped unprefixed, but this should not be a problem because > current > > browser properly ignore those declarations, just like they ignore unknown > > properties. > > The fact that they currently ignore them actually means that we > probably can't change our behavior - people may have left properties > in their animations and are now accidentally depending on them not > working. > > However, the use-case is very valid. I think we should be able to > solve this with a new timing function that works for all properties > (and define that the others only work for "animatable" values) - > discrete(<percentage>). The percentage indicates at what point in the > time it should flip from the start value to the end value. > It would be terrific if we could make this happen. A side issue is that animation are specified to not run if they have 'display:none'. So, as soon at the rule switches the display property, the animation would stop running which is not what the author intended. I have no idea how difficult it would be to implement this...
Received on Wednesday, 1 August 2012 19:36:53 UTC