On Mon, Dec 06 2010, Lars Magne Ingebrigtsen wrote:

> The point of doing it this way is that if you have
>
> <p style="color: red">foo<p>bar<p>zot
>
> then each of these will apply "red" to their own bits.  That's slow an
> unnecessary, since, just having the top-level <p> do it will yield
> exactly the same result, won't it?

In that example yes. But with:

<p style="background-color: red">foo<p style="color: red">bar<p>zot

The result should be quite different.

With the previous code, you would:
- Insert `foo'
- Insert `bar'
- Paint `bar' in red
- Insert zot
- Paint the backgruond for foobarzot in red.

bar was invisible: red on red. The color checker had no clue that the
background was red.

Now it does:

- Insert `foo'
- Enhance the style of the next p with "background-color: red"
- Insert `bar'
- Paint `bar' in red with a red background… Wait? No the color check
  steps in and choose better color in the rainbow just for you! HAHA!
- Insert `zot'
- Paint foobarzot with a red background, but not bar, since
  shr-put-color have some magic to not override color information IIUC.

Sounds good?

-- 
Julien Danjou
// ᐰ <julien@danjou.info>   http://julien.danjou.info