- From: Alan Stearns <stearns@adobe.com>
- Date: Tue, 1 Oct 2013 17:08:46 -0700
- To: fantasai <fantasai.lists@inkedblade.net>, "www-style@w3.org" <www-style@w3.org>
On 9/30/13 11:57 PM, "fantasai" <fantasai.lists@inkedblade.net> wrote: >2.2. Basic Shapes > > # When using this syntax to define shapes, the relevant > # box is determined by the computed value of the > # Œbox-sizing¹ property. > >I'm not sure that this makes a whole lot of sense, given >that by default floats use the margin box as their float >area, and by default box-sizing is 'content-box'. > >I don't have a solution in mind, I just think this is wrong. >I think rectangle(0,0,100%,100%) should be equivalent to the >default behavior; that it's not, strikes me as inconsistent. My first inclination was also to have that rectangle be equivalent to the default behavior, but there is value in being able to select different box coordinates (particularly for using the same shape on elements with varying borders). We got several requests to be able to choose the relevant box, and exactly no requests to be able to simulate the default behavior with a basic shape. I think setting 'shape-outside:auto' is sufficient for the default behavior. > > # The coordinate system for the shape has its origin on > # the top-left corner of the relevant box with the x-axis > # running to the right and the y-axis running downwards. > >This could present some problems... if you're defining a >contour that overlays an image, it's exactly what's needed, >but if the shape is relating to the arrangement of content >on the page, you might want it to be logical, like the grid >coordinates, rather than physical. > >Again, I don't have a solution in mind. :( But I think we >need to have one in mind. We are following the clip:rect() example of only using physical coordinates, mainly because that was the precedent we had to start with. Since <basic-shape> parameters are all defined in terms of widths and heights of the box, which are physical dimensions, it makes sense to use physical coordinates. Using only physical dimensions has worked very well in the implementation, and we have tons of writing mode tests. I think we'd need to see a use case to justify the complexity of defining how <basic-shape>s could flip to using logical coordinates. Thanks, Alan
Received on Wednesday, 2 October 2013 00:09:13 UTC