Re: Comments on gradients proposal

On Thu, Oct 22, 2009 at 4:51 PM, Robert O'Callahan <robert@ocallahan.org> wrote:
> The processing of gradient stops could stand some additional clarification.
> I think the model is:
> 1) any stop with a specified position that is less than the specified
> position of any stop before it has its position changed to the greatest
> specified position of any stop before it.
> 2) *then* you interpolate the stop positions of any stops with unspecified
> positions
> 3) for radial gradients, drawn circles/ellipses with negative radii are
> ignored (negative stop values are not changed). (There may be massaging of
> stops before passing to native APIs, but this is purely an implementation
> detail.) This is more important if you add support for repeating gradients,
> for example it should be clear that repeating-radial-gradient(white -50px,
> black -25px, white 0) doesn't just set all the stops to position 0 and
> render white.
>
> I think you can deduce all that from the spec, except possibly the ordering
> of steps 1) and 2), but it might be clearer if it was in one place.

Correct on all counts, and I'm making changes to the language now.

~TJ