Yeah, that’s a classic issue when you’re relying on floating point arithmetic or integer rounding. Even a tiny off‑by‑one error can propagate through recursion and throw the symmetry off. One trick is to keep everything in integers—like using a fixed‑point system or scaling your coordinates by a large factor—then round only at the very end. If you’re drawing on a canvas, you can also snap the final line endpoints to the nearest pixel grid. Another angle is to compute the whole shape first, then adjust the final vertices to enforce perfect alignment, basically solving a small system of equations to enforce the constraints. Did you try isolating one level of recursion and rendering it separately to see where the deviation starts?

Sure, just make a small helper that logs the coordinates at each depth. For example, at the start of your recursive function, push the current depth and the endpoint pair into a flat array. After the recursion finishes, iterate over that array depth‑by‑depth and check if the endpoints match the expected values you’ve pre‑computed. If you see a discrepancy, it’ll be obvious at which depth it diverges. Also, keep the comparison tolerance tight—maybe 0 or 1 pixel—and print both the expected and actual coordinates. That way you can pinpoint the exact recursion level where the alignment breaks. Good luck!

Sounds good, keep an eye on that log and let me know if the error turns up at a specific depth. I’ll be here if you hit another snag.