change list for WebCore/JavaScriptCore version 60

Dirk Mueller mueller@kde.org
Sat, 15 Feb 2003 02:40:47 +0100


On Fre, 14 Feb 2003, David Hyatt wrote:

> 1. TextSlaves paint the backgrounds/borders/outlines of their parent 
> inlines.  This means you can end up painting in the wrong places (or 
> not painting at all) when another inline gets in the way, e.g., 
> <a>First <i>Second</i> Third</a>.
> 2. Text-decoration is painted by the text runs, when it should be drawn 
> by the parent inline element (and able to extend under/through elements 
> like images and form controls.

Yeah, that sounds familiar, because thats exactly what I'm working on 
currently. 

do you have some code to share already? I've made some experiments but no 
concrete idea on the best implementation for this yet. We've tried hard to 
avoid giving the CSS line boxes an actual representation in the rendertree 
so far, and I hope we can keep it this way.  

> 3. Space isn't left for descent on lines with no text (e.g., 
> <td><img></td>).  In strict mode that space should be left (this issue 
> is the reason "almost strict" mode needs to exist).

ugh. do we have to implement this detail? anything major website that 
breaks?

> 4. There are numerous issues with :first-line as well.

Yes. don't forget to look at the code in HEAD, I remember a few 
related bugfixes. 

> Basically the code I'm working on now makes boxes for inlines (just as 
> RenderText makes TextSlaves for its runs).  The inline boxes are doing 
> their own painting of their borders/backgrounds and text decorations.

I'm not sure I understand. All inlines inherit currently RenderBox, so it 
just requires storing the static position offsets during bidi positioning 
in the box. 

Is this what you mean?

The problem is this doesn't work for background-positioning of nested 
inlines. I've no solution for that yet.


-- 
Dirk (received 126 mails today)