Re: [CSS3] linear-gradient() proposal

On Sat, Mar 26, 2011 at 11:49 PM, Rik Cabanier <cabanier@gmail.com> wrote:
> The decision to not have a vector to describe the gradient is the reason you
> run into this problem.
> In the past I brought up that having a transition on the angle will produce
> a 'wobble' effect on the gradient's look because the length of the gradient
> will vary during the rotation.
> The current design was deemed simpler than a vector so it wasn't changed.

If I recall correctly (I haven't gone back and checked the archives,
so I might not be recalling correctly), your "wobble" feedback was
about the apparent speed of rotation if the gradient was envisioned as
a vector and you did simple linear interpolation on the position.  I
took that into account in the latest revision, so that the pure-angle
interpretation of vectors is the "correct" one, and interpolation
always happens over the angle so you get smoother transitions.

The "length" of the gradient does change if the gradient is
transitioned to a different angle, but I think that's desirable in at
least some cases, and in the cases where it's not, you can usually get
around that just be specifying the color-stop offsets using lengths
instead of percentages.  The cases that are left unsatisfied are, I
believe, less of a problem than the rotation speed "wobble" that is
solved by this interpretation.

It's unfortunately impossible to satisfy every case without overly
complicating the syntax; we could offer more in languages like
Javascript or SVG that can absorb the complexity, but CSS is supposed
to be as simple as possible to use.

~TJ

Received on Monday, 28 March 2011 01:43:14 UTC