Re: [css3-transitions] Back-tracking transition-timing-function

On Apr 22, 2010, at 5:29 PM, Tab Atkins Jr. wrote:

> On Thu, Apr 22, 2010 at 5:23 PM, Alex Meiburg <timeroot.alex@gmail.com> wrote:
>> In the case that the points P1 and P2 were too over extended from (0,0) and
>> (1,1), respectively, the curve they generate could end up back-tracking to
>> some extent. In the event of some bad code that does this, how should the
>> user agent handle it? Should it simply skip from the lower branch of the
>> function to the higher branch? Should it readjust the values of P1 and P2 so
>> that the back-tracking is removed? Should it ignore the transition entirely?
>> 
>> In addition, should coordinates of P1 and P2 outside the [0-1] range be
>> allowed? There are some well formed cubic bezier curves with control points
>> outside the square that stay inside the square continuously.
> 
> The cubic-bezier() function only allows values in [0,1].  Any values
> outside of that range make the declaration invalid.
> 
> Within that range, you're guaranteed that the curve will be
> monotonically increasing (or maybe just non-decreasing?), and thus is
> invertible.

It's true that this is what the spec says. We did a functional notation for timing-function to allow for more complex curves in the future. But for now I think the existing features are sufficient. You can get more complex timing functions with animation keyframes. Not as simple, but possible.

-----
~Chris
cmarrin@apple.com