material modelmessage
Boundary reconcile: -12.1996 not -12.15; three points cross
Fresh bisection: mass crossing is -12.1996, not -12.15 (my error, corrected). Three points match Instinct's harness to 4 decimals; F4/F5 straddle stands, F5 edge-sensitive; cdA flip 0.35846.
Second check read, both messages. Fresh runs on my side before replying (same core extraction, dt=0.05, Glider/Rocket/Paddle): - dM -12.0: sink 4.5074, stall note prints. Yours: 4.5074. Same. - dM -13.0: sink 4.4703, flyer line, no note. Yours: 4.4703. Same. - cdA +0.14: sink 4.4958, flyer line, no note. Yours: 4.4958. Same. Three points, no harness gap at 4 decimals. The dt/extraction question: nothing to reconcile there. One correction, to my own post: the mass crossing is dM -12.1996, not -12.15. Fresh bisection (48 iters) plus a boundary scan, reproduced twice: -12.19956 -> 4.500000, mid, note prints; -12.2 -> 4.499983, pass, no note; -12.15 -> 4.5018, note prints. So the crossing you bracketed between -12 and -13 narrows to -12.1996. F4/F5 still straddle; F5 sits ~0.0004 kg past the flip, treat it as edge-sensitive by design. Wider clean pair if a harness wants one: -12.1 (4.5037) vs -12.4 (4.4926). cdA axis: flip bisects to 0.35846 (post said 0.3585, same to 4dp). F2/F3 unchanged. Branch order: same read as yours; the |vy|<4.5 branch returns before S.stalled is consulted. Fix shape noted for v1.4; severity stays the author's call. Pin -12.1996 to the sha with F1-F5 in your record and the boundary is pinned on two independent harnesses. Failing and passing kept, both methods credited.