- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Tue, 7 Sep 2010 14:20:25 -0700
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: "www-style@w3.org" <www-style@w3.org>

On Tue, Sep 7, 2010 at 1:50 PM, fantasai <fantasai.lists@inkedblade.net> wrote: > We seem to have multiple notions of how angles should work. > > There are two commonly-taught methods of mapping angles to directions: > - the Bearing Angle Method > - the Cartesian Angle Method > There are other methods of mapping angles to directions: > - the Graphical Angle Method > > Bearing Angle Method: > - Used for compass bearings > - Zero degrees points north/up > - Angles increase clockwise > - Used by the following properties: > azimuth (CSS2) > glyph-orientation (SVG, XSL) > image-orientation (Paged Media) > > Cartesian Angle Method: > - Determined by arctan(slope) in a Cartesian coordinate system with > x-values increasing rightwards and y-values increasing upwards > - Zero degrees points right > - Angles increase counter-clockwise > - Used by the following: > CSS gradient notation (implied by illustrations in CSS3 Images) > > Graphical Angle Method > - Determined by arctan(slope) in a graphics coordinate system with > x-values increasing rightwards and y-values increasing downwards > - Zero degrees points right > - Angles increase clockwise > - Used implicity by the following: > Scalable Vector Graphics (SVG) > CSS 2D Transforms > > Of the two that fall under the Graphical Angle method, afaict neither > uses absolute angles -- the zero point doesn't matter, only the clockwise > increase. Which makes SVG and Transforms compatible with the Bearing Angle > Method from a user perspective. > > This makes the gradient syntax the odd one out. Therefore its use of angles > should be > a) defined in the prose somewhere instead of implied by illustration > b) use the Bearing Angle Method, i.e. 0deg points up and angles increase > clockwise > > Furthermore, the CSS3 Values and Units module should make this convention > explicit so that later spec-writers don't make the same mistake. (Also, > the outdated reference to ACSS and the 0-360 normalization requirement > should be removed.) Argh, I was hoping I could avoid this, but it seems like I maybe can't. ;_; Anybody have any strong objection to me switching the <angle> reference to Bearing Angles? ~TJ

Received on Tuesday, 7 September 2010 21:21:18 UTC