- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Tue, 15 Oct 2013 10:38:00 -0700
- To: Shane Stephens <shans@google.com>
- Cc: "public-fx@w3.org" <public-fx@w3.org>
On Sun, Oct 13, 2013 at 8:30 PM, Shane Stephens <shans@google.com> wrote: > What should the result of interpolating (e.g. for clip): > rect(0px, 0px, 0px, auto) > to > rect(100px, 100px, 100px, 50px) > be? Ooh, this is a Transitions bug - it defines clip to interpolate as "rectangle", which assumes that all of the values are numbers. (Which is even more wrong - they should interpolate as lengths.) The obvious answer is that these two aren't interpolable (at computed-value time), so the rectangle should just flip. > What about (e.g. for mask-box-image-slice): > 50% fill > to > 30% This needs to be defined by Masking. Any property whose value is non-trivial needs an Animatable line in its propdef table. I suspect the obvious answer is that it should interpolate from 50% to 30%, and flip from filling to not filling, but I could see arguments that it should just flip the entire thing. > Should individual longhand properties with multiple 'channels' interpolate > each channel separately (i.e. 0px can smoothly interpolate to 100px even > while auto steps to 50px), or should a single non-interpolating channel > block interpolation of all channels (i.e. fill -> not fill is a step, so 50% > to 30% should step too)? Defined by the longhand. There's no general principle, as different things make sense. For example, shadows are only interpolable if both are inset or both are outset. ~TJ
Received on Tuesday, 15 October 2013 17:38:47 UTC