Given this HTML:
<div sink=container>
before
<!start>
between
<!end>
after
</div>The parsed result would have 3 child text nodes. The API would behave like:
div.beforeSinkStartis the "before" text nodediv.afterSinkEndis the "end" text nodediv.getSinkRange()(if it exists) would return a range spanning just the "between" text node
Serializing would have to add <!start> and <!end> between the appropriate text nodes.
If there's anything that automaticaly merges text nodes (like div.normalize()) we might also need to prevent that.