More importantly than being able to set
innerHTML on DocumentFragment, to me, would be an ability to create a fragment pre-filled with a provided array of child elements, like
What about evaluating it (when set) as if it were in the root context of a fresh document (with the elements being appended)? For reasonably well-formed HTML (ie. not omitting contextual end tags), what problems would arise? Are there elements that cannot be parsed within the root context?