make pseudo-elements non-anonymous
David Hyatt
hyatt at apple.com
Tue Mar 2 22:49:55 CET 2004
Consider this example:
<head>
<style>
p:first-letter { font-size: 150%; float: left; border: 3px solid green;
padding: 5px; }
</style>
</head>
<body>
<p><a href="http://www.apple.com">Click here</a> to visit apple.com
etc. etc.</p>
If the "C" first-letter container has <p> as its DOM element, then the
DOM will deliver mouse events to the <p> when you're over the
border/padding area of the first-letter, bypassing the anchor entirely.
Of course this also doesn't work right now (with the first-letter
container being anonymous), but it feels conceptually cleaner to me to
leave the generated content container for first-letter without a DOM
element.
For before/after it seems ok to use the element(), although if you're
only doing it for the container (and not also for the anonymous
children), then I'm not sure what the benefit is... i.e., we still have
to fix the problem of obtaining the innermost DOM node that is the
ancestor of some anonymous child that has no DOM node (and doing that
would fix all of these problems).
dave
On Mar 2, 2004, at 7:07 AM, Germain Garand wrote:
>
> Hello,
> As pseudo-elements are, from a logical standpoint, an intrinsic
> extension of a
> given container and as such not really anonymous, we'd like to make
> them
> share said container's element().
>
> This has the benefit of avoiding the need for further arbitrary
> discriminations amongst anonymous blocks eligible for squeezing, and
> to keep
> the relationship with said container clear in the render tree, even
> when the
> pseudo-container isn't a direct child (case of first-letter).
>
> Is this something that could be shared accross our trees ( e.g
> simplifying
> your isAnonymousBlock method into something that tests what it says it
> test)
> or do you have a better idea?
>
> Greetings,
> Germain
> <pseudo.diff>_______________________________________________
> Khtml-devel at kde.org
> https://mail.kde.org/mailman/listinfo/khtml-devel
More information about the Khtml-devel
mailing list