Re: [css3-images] linear-gradient keywords and angles are opposite


I disagree that the vector syntax is more complex than the current one.
This whole email thread is proof that the spec is not that easy (and it is
not the first time that this has been mentioned). Maybe it's easier for the
simple cases, but it quickly digressses when you try to do something more
complex or when you explore edge cases.

I also disagree that angles are how gradients are drawn in design software.
In all Adobe software, gradients are defined with the gradient tool.
(Mousedown = start of vector, MouseUp = end of vector)
Recent versions of Illustrator even show the vector when you select the
gradient and let you modify its control points. Illustrator lets you change
the angle but that is not how people draw in general.

The underlying drawing models as well as PDF and PostScript also define
axial and radial gradients on a vector.

The fact that you can draw stars with the current syntax is not relevant
(and could be done with the vector syntax too...)


On Thu, Jun 9, 2011 at 9:37 PM, Brad Kemper <> wrote:

> On Jun 9, 2011, at 9:06 PM, Rik Cabanier <> wrote:
> > I completely agree.
> > I made a similar proposal last December but was told the spec was too far
> along.
> That wasn't the only reason. But it had been debated a lot before that.
> > A vector is much easier to define which probably why it is used in most
> design software.
> All the design software I have ever seen that lets you specify direction by
> typing in text, does so via angles. Angles are simpler and easier for humans
> to read and immediately suss the direction (you just have to know where zero
> is, and which was the degrees progress around a reference circle).
> The extra power of being able to start or stop the gradient at arbitrary
> points is already present to a huge degree with the color stops. The cases
> where that isn't quite enough are few and far between, and still doable via
> SVG. I'm all for simplicity to serve the vast need,s and not adding extra
> syntax that adds very little extra expressiveness for a few rare edge cases.
> If you look at some of those examples that Tab posted a while back, with
> stars and plaids and such being created via gradients, I'm extra happy that
> Tab kept the syntax limited. It gets hard enough to read once there are
> multiple backgrounds, each with multiple color stops and precisely placed
> color stop locations. Layering bg-position syntax within that, while there
> might also be actual background position for the tiling of the gradients,
> and you end up with a mire of numbers that are slow to comprehend. I'd liken
> it to multiplying GPS coordinates together to try to imagine where something
> is in the world.

Received on Friday, 10 June 2011 05:13:31 UTC