d12459d1e1
nav: lane DN, the instrument -- what the classifier hook declares and where each input comes from
586171cb72
app: T34 RecordObservedDesigns -- the observed-design list, and the three rules the save cannot show
feaa70e15e
lane DT merged: T34 implemented, 36 closed / 2 regressed on the rich turn; the other 24 leaves have two named dependencies
d9b06ca447
lane DT: T34 implemented and measured -- 36 closed / 2 regressed on the rich turn, 1 / 0 on the canonical pair
fdea0fde08
lanes CR and DW merged: research stays compared (16 leaves, all OnTechResearched); writer byte-exact on 43 saves; gate locked; exhaustive-config blind spot
7dcc66bc66
lane CR: our code ran instead of ProcessResearch on a real completion; the oracle missed by 16 leaves, all of them OnTechResearched's
ea7b771c76
lane DT: T34 RecordObservedDesigns -- mechanism read from the instruction stream, prediction committed before the build
b6018a0f19
plan: pivot to displacement; gate.sh replaces the hollow gate; Rung B worklist from lane CV
f1306713d1
lane BT merged: --relabel-new-ids ships with both acceptance halves and five refusing guards
d6f725a7ae
lane BU merged: H1 confirmed, the visit key is fleet->Location, the outcome set is two heap pointers
696935ea2a
lane BU: H1 confirmed - the fleet-visit order is ascending fleet->Location, and the two-file outcome set is the two orderings of two heap pointers
232550595d
lane BU: run 1's result reshapes the prediction; the run-2 test committed before run 2 finished
51243ef8b3
lane BS merged: the per-history residue is an absent Colonize task; BR's generator row scoped to encounter-free turns