Re: Linear gradients, Transforms and angles...

On Sep 20, 2010, at 8:05 AM, Tab Atkins Jr. wrote:

> The difference was unintentional, in that I didn't *mean* for it to be
> different from other properties.  But I did specially intend for it to
> be as it is, with 0deg pointing East and CCW being positive.  That's
> how polar angles work, which I treated as the most common case of
> directed angles.

I think that when rotating something, it it natural to start at a default angle (0°, pointing to the right), and rotate it a default direction (clockwise), and so that is what transforms, people who program SVG applications, etc. have.  But for linear direction, you aren't rotating anything, from most people's point of view. You are picking a single direction, and the angle you think of it as, commonly, is the one that is printed on protractors and used in geometry textbooks, where 90° is "up" (and 0° is to the right).

Consider turns. You could specify the gradient direction in turns instead of angle, right? But that is not intuitive, because an author is not going to think of a gradient direction as turning anything. With turns, the direction is clockwise. With linear gradients, the direction is, well, linear (like "North" is a linear direction, not a rotation). I don't mean we have to disallow turns, just that picking a direction is not intuitively the same as rotating something, and uses a different convention for thinking about where 90° is.

Also, in design software, it is more common to see 90° as up, as it is in the market-leading Adobe CS software (PhotoShop, Illustrator, etc.).

Received on Monday, 20 September 2010 16:09:49 UTC