but i think the difference is that in the v2 engine we don't recreate the parse context for every target