Is the garbage collection in picolisp immediate if the resource can be released?

The initial iterative replacement solution I came up with (below) quickly runs out of resources as it has to build the entire list before starting to draw. I haven't figured out how to do it yet, but I think, in this case, a 'recursive' solution that renders as points are created would use a lot less resources, assuming that processed replacement rules are garbage collected as the stack unwinds and elements are traversed over. It would be a nice accomplishment to be able to render the more complex fractal plants expressed here: https://en.wikipedia.org/wiki/L-system /Lindsay # https://en.wikipedia.org/wiki/Gosper_curve # Angle 60 degrees = PI/3 # Axiom A # Replacement Rules # A :--> A - B - - B + A + + A A + B - # B :--> + A - B B - - B - A + + A + B # C (initially = A) grows...very quickly. #: n=0 (length C) -> 15 #: n=1 (length (setq C (F C))) -> 113 #: n=2 -> 799, n=3 -> 5601, n=4 -> 39215, n=5 -> 274513, n=6 -> 1921599, n=7 . # For n=6... : (bench (nil (Draw-Gosper-Curve))) -> 1.370 sec Which works out to: XY Cnt: 823543 coordinates to plot Min-XY: -66,432.00 -42,345.11 Max-XY: 2,048.00 23,611.35" ..about 900K of serialized canvas instructions. Full-code at: https://github.com/thinknlive/picolisp-gosper (de Gosper-Curve (Length Angle N) (let (A (chop "A-B--B+A++AA+B-") B (chop "+A-BB--B-A++A+B") C A F '((L) (fish atom (mapcar '((X) (cond ((= X "A") A) ((= X "B") B) (T X))) L))) ) # Generate points (do N (setq C (F C))) # Plot points (map '((R) (case (car R) (("A" "B") (Plot-Line Length Angle)) ("+" (setq Angle (+ Angle PI/3))) ("-" (setq Angle (- Angle PI/3))) (T (msg (text "?Gosper-Curve: No-match: @1" @))) )) C ) ) )