This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
(This bug was originally https://www.w3.org/Bugs/Public/show_bug.cgi?id=22898) In the HTML spec, any node that is not an ancestor or descendant of the active modal dialog is marked inert.[1] It seems the ancestor/descendant relation should be based on the composed tree (a Shadow DOM concept). Otherwise, children that have been transposed into a modal <dialog>'s subtree are considered inert, although they are rendered as if they are real children of the <dialog>. This is surprising to the user interacting with the dialog. I guess to spec the desired behavior, the HTML spec will need to make a mention of the Shadow DOM spec. See also https://code.google.com/p/chromium/issues/detail?id=269842 [1] http://www.whatwg.org/specs/web-apps/current-work/multipage/editing.html#inert
The way I did this in XBL, IIRC, was to define a merged tree that CSS works on, and to say that certain algorithms work on that tree instead of the "real" tree. I'd rather not have to mention shadow trees in all the relevant algorithms in HTML, at least not until the Shadow DOM stuff is far more established (at which time maybe we would just merge it into the DOM spec and update HTML accordingly).
Matt, is this something you'd be willing to patch in HTML?
Moved to https://github.com/w3c/webcomponents/issues/653