On Thu, 9 Jan 2025 21:41:09 GMT, John Hendrikx <jhendr...@openjdk.org> wrote:

> I suspect browsers have an easy way to track the index of each child (without 
> having to renumber them). All that would take is a different underlying list 
> type as Nodes must be unique. A Tree structure for example can determine the 
> index of a Node with O(log N) performance for adds/removals/find index. If 
> browsers support advanced features like `nth-child(2n+1)` I suspect they will 
> not be using a structure that require renumbering all items or require a full 
> search to determine the index.

Yes, the problem will be to reconcile that with the existing API around 
pseudo-classes. It's obviously the case that a node will match an unlimited 
number of functional pseudo-classes like `nth-child()`, so we can't enumerate 
them like we do now.

-------------

PR Comment: https://git.openjdk.org/jfx/pull/1652#issuecomment-2581307009

Reply via email to