I didn't know about #2, that's really cool. How does render_to_string handle cached partials? Does it still need to compute the output of the partial's string

I didn't know about #2, that's really cool. How does render_to_string handle cached partials? Does it still need to compute the output of the partial's string representation on every request?