With the spa_auto support this becomes feasable and avoids problems when the pod size is unknown and ends up larger than our stack buffer.