Your principle holds: it's always better, when exploring a new field, to start without try the most minimal libraries and frameworks after you already experience the pain they solve.