Re: [css3-transform] definition of skewing

On Jan 12, 2011, at 8:52 PM, Rik Cabanier wrote:

> All,
>  
> It seems that the definition of skew in the CSS transform is a little odd.
> http://www.w3.org/TR/SVG/coords.html#SkewXDefined contains the definition of skewx and skewy. As you can see, the calculation is using ‘tan(a)’.
> This means that is you have a skew from 45 to 135 degrees, the animated content will stretch as well as skew.
> It seems that this is not really the desired behavior…
>  
> If it is, I would like to propose that we extend the ‘rotate’ transform to take an additional parameter.
> Ie rotate(45deg, 55deg) would rotate the x-axies by 45 degrees and the y-axis by 55.
> I think that is a more natural use of skew.

We used to have skew(x, y), but see this message and followups on why that was removed:
<http://lists.w3.org/Archives/Public/www-style/2009Mar/0287.html>

I object to re-using rotate() to do this kind of skewing. It's really a different operation to rotation.

skew(angle) is well-defined by SVG <http://www.w3.org/TR/SVG/coords.html#EstablishingANewUserSpace> and we follow that definition for CSS.

Simon