- From: Anne van Kesteren <annevk@opera.com>
- Date: Tue, 27 Oct 2009 16:31:50 +0100
- To: "Xiaomei Ji" <xji@chromium.org>, "Maciej Stachowiak" <mjs@apple.com>
- Cc: Olli@pettay.fi, public-webapps@w3.org, sam@webkit.org, webkit-dev@lists.webkit.org
On Fri, 16 Oct 2009 20:20:00 +0200, Xiaomei Ji <xji@chromium.org> wrote: > 1. Why we would like to differentiate whether the node is a 'document' > node or a 'control' node? Can such differentiation be achived by > existing flags in node instead of "offsetKind"? Yes. I think we should nuke offsetKind for now. > 2. If we expose the control node, how to deal with the concerns Olli > raised? > "Do you really want to expose those (native) anonymous DOM nodes to > web? What should happen if one tries to append them to normal DOM? > Or removes them? Or adds (mutation) event listener to them?" You get a numeric offset, not any kind of object. > 3. From user's point of view, I think we probably need to have a > convenience > method for converting a Range, and it probably should not give null for a > control position. We would need both the node and the offset for further > processing. If we have a convenience access to a Range it would have to be null for replaced elements that contain anonymous line boxes. Is that really needed though? -- Anne van Kesteren http://annevankesteren.nl/
Received on Tuesday, 27 October 2009 15:32:40 UTC