- From: Zack Weinberg <zweinberg@mozilla.com>
- Date: Thu, 19 Feb 2009 12:03:06 -0800
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: W3C Emailing list for WWW Style <www-style@w3.org>
fantasai <fantasai.lists@inkedblade.net> wrote: > Zack Weinberg wrote: > > We noticed while reviewing internal test cases for border-radius > > that Gecko permits <percentage> values for all border-radius > > properties, contra the spec which allows only <length>...What do > > y'all think of adding this extension to the spec? As far as I can > > tell, no other implementation has this already, but it should not > > pose any serious difficulty either in the parser or the renderer. > > So, one issue there is that in Gecko's original version, there was > only one radius. Your curves would always be a quarter-circle. In > CSS3, you have two. You can have elliptical curves. Instead of > percentages always referring to the width, you could also have > percentages referring to their respective dimension of the box. > We weren't sure what to do here, so we left that out. Yeah. I can argue either way, but on balance I prefer "always width" because that way the behavior of "border-radius: 10%" with a rectangular box is to give quarter-circles, consistent with "border-radius: 10pt". Also, the height of a box tends to be much more variable than its width, so designers are more likely to want to style relative to the width, I think. > Another issue > in the decision was that if in the future we wanted to allow content > to respect the curve, it'd introduce a circular dependency. It would just be a system of simultaneous constraints, wouldn't it? zw
Received on Thursday, 19 February 2009 20:03:50 UTC