Re: [CSSOM] Interaction of getBoundingClientRect/getClientRects with transforms

On Wed, Sep 22, 2010 at 4:21 AM, Boris Zbarsky <bzbarsky@mit.edu> wrote:

> On 9/21/10 2:17 PM, Simon Fraser wrote:
>
>> I agree. I'd expect one of two behaviors for rotated inline elements. The
>> first would be "slab" rotation, where all parts of the element are rotated
>> on a plane (with some center point that we'd specify, probably the center of
>> the bounding box of the parts of the span). The second would be that the
>> rotation is applied per line, somehow.
>>
>
> Both of these really do look pretty weird with bidi text (for example, it's
> pretty easy to end up with a situation where the boxes on a line are:
>
>  XXXX YYYYYYYYYYYYYYYYYYY XXXX
>
> where the Xs belong to one span and the Ys to another.  When applying a
> transform to the X span, things would look pretty odd.
>
> Maybe that's just a cost of supporting both bidi reordering across inline
> element boundaries and sufficiently rich styling of inline elements.  I
> guess borders look pretty weird too, in that situation.


That certainly would look weird, but I don't think it's any weirder than the
current Gecko behavior. More importantly, I think web developers would
understand what was going on here (to the extent that anyone understands
bidi!) and not have difficulty coming up with a workaround.

Ojan