StoneCypher / jssm / 47e2e675272d7b54ac3a01df1f104cc3939f5fd8 / 5

100%

master: 100%

Pull #634

github

test(bench): wire benny:scaling/benny:all and capture v5.128.0 baseline

Diagnostic signals from scaling.md:
1. chain transition() across N=10..1000: not flat, drops ~45% (137705 -> 75764) across a 100x size growth, i.e. sub-linear but visibly degrading
2. dense transition() vs chain at same N: dense degrades faster but only modestly - dense-10 is ~25% slower than chain-10, dense-200 ~36% slower than chain-200; not a sharp collapse
3. edges_between() scaling with edge count: yes, scales with edge count - chain-1000 (~1000 edges) ~1212 ops/s, dense-200 (~40k edges) only 11 ops/s, messy-5000 only 21 ops/s
4. hooked-200 vs hub-200 transition(): yes, hooked-200 (10624) is ~6.5x slower than hub-200 (68880) - hooks dominate per-transition cost
5. has_state() flatness: yes, flat across all sizes (~9.6M to ~14.4M ops/s, noise band) regardless of shape or N
6. messy-1000 vs chain-1000 per op: yes, perceptibly worse on most ops - transition 60156 vs 75764, edges_between 307 vs 1212 (~4x slower), construct 13 vs 93 (~7x slower); has_state alone is comparable

938 of 938 branches covered (100.0%)

Branch coverage included in aggregate %.

1504 of 1504 relevant lines covered (100.0%)

1881.17 hits per line