Re: [css-shapes] interpolation update - ready for LC again?

On Fri, Nov 22, 2013 at 7:03 AM, Alan Stearns <stearns@adobe.com> wrote:
> On 11/22/13 5:50 AM, "Dirk Schulze" <dschulze@adobe.com> wrote:
>
>>
>>> On Nov 22, 2013, at 1:28 AM, "Alan Stearns" <stearns@adobe.com> wrote:
>>>
>>> Hey all,
>>>
>>> I've updated the interpolation section [1] of CSS Shapes and added
>>> 'Animatable' lines to the propdef tables. This is the last of the
>>>changes
>>> I think were required from TPAC decisions, so I think the draft is ready
>>> for last call again.
>>>
>>> In summary, if you're interpolating between <basic-shape> functions, the
>>> functions need to be the same shape, use the same reference box, and
>>>avoid
>>> keywords that have no interpolation defined. If all of these strictures
>>> apply, then you interpolate between the function parameters as a simple
>>> list of length, percentage or calc. But I've added one additional rule
>>> that allows interpolation between identical keywords. So 'circle(3em at
>>> top left)' can interpolate with 'circle(6em at top left)'
>>>
>>
>>Why not define interpolation for different parts of a shape? For circle
>>you can split it in interpolation for radius and interpolation of
>>position. While one may not be animate able, the other still is. Or is it
>>that what you did?
>
> I did not do that. I just made it so that you can interpolate between
> identical keywords.
>
> When a property is defined as interpolating as a simple list of length,
> percentage or calc, and the two lists are:
>
> 10px foo 10%
> 20px bar 20%
>
> Does it mean that the property interpolates 10px to 20px, and 10% to 20%,
> and foo to bar is ignored? If that's the case then my additional rule
> about identical keywords isn't needed.

No, if it's defined as interpolating that way, but the value isn't "a
list of length, percentage, or calc", then there's a spec coordination
problem.

In particular, background-position interpolation is defined wrong
because of this.

~TJ