Re: [css3-images] processing model for transformations (image() vs. image-orientation)

(Cc+ public-i18n-bidi)

(11/12/01 8:22), Tab Atkins Jr. wrote:
> On Wed, Nov 30, 2011 at 4:03 PM, L. David Baron <dbaron@dbaron.org> wrote:
>> There are multiple features in css3-images that can transform an
>> image.  For example:
>>
>> http://dev.w3.org/csswg/css3-images/#image-notation defines the
>> 'ltr' and 'rtl' keywords which can cause an image to be flipped
>> horizontally.
>>
>> http://dev.w3.org/csswg/css3-images/#image-orientation defines the
>> image-orientation property which can cause an image to be rotated.
>>
>> It matters which one happens first, since doing them in a different
>> order produces different results when image-orientation computes to
>> 90deg or 270deg.  css3-images should define a processing model that
>> says so.
> [snip]
>
> Though, I guess they could potentially interact once 'content' can
> define replaced elements.  In that case, I think 'image-orientation'
> should apply *after* the contents have resolved to an image.  I'll add
> some text to that effect.

The spec now says

  # If the image itself is transformed in some way (for example, if the
  # content of an element is provided by the ‘image()’ function with a
  # directionality opposite the element's directionality), the image's
  # transformation must be applied before ‘image-orientation’ is. As
  # well, ‘image-orientation’ must be applied before any further
  # transformation of the element, such as through CSS Transforms.

Is this actually right? I thought 'image-orientation' is meant to allow
you to "correct" the image, say, from 「↑」 to「→」, so it should happen
before anything else. If ltr/rtl reflection happens before 'image-
orientation'（「↑」to 「↑」）, you get only「→」instead of both 「→」and
「←」.

Am I missing something?

Cheers,
Kenny