[Webkit-unassigned] [Bug 36724] Add support for <ol reversed>

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=36724





--- Comment #7 from Darin Adler <darin at apple.com>  2010-03-29 15:25:55 PST ---
(In reply to comment #4)
> I'm not sure how to fix that. Because that basically means listening for
> changes in the entire subtree or throwing the whole optimization away (which I
> don't think is a particularly good idea). And I wonder if it's even worth it
> (it's a broken HTML code anyway).

We need correct behavior even with "broken" HTML. Dynamic changes should not
give different results than the same markup in non-dynamic form. I think we do
need to do the equivalent of listening to changes in the entire subtree. Maybe
we can find an efficient way to do it. I suspect we can.

> I went for that and added a function to the LayoutTestController that returns
> an array of item markers for a given list and rewrote the tests to plain text.

I was thinking instead that the function would take a node and return the list
item marker text for the renderer corresponding to that node, or null if there
is no list item.

> While it's definitely better to share the same test result across ports, in
> this case the tests are kinda less readable, at least in its current shape.

Maybe I can help you improve the tests. I was thinking we could use the
function I had in mind and a little JavaScript code to generate a nice,
readable outline in plain text.

> Also, the added function needs to know the logic of how list elements and item
> elements interact with each other, which makes it more error-prone and if it
> has bugs itself, it can hide other bugs in the implementation.

That's surprising. I suggest we find a way to do it and use the
RenderListItem::markerText() function.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.