v1.4.0¶

Features¶

bench: widen the data-packet predicate to accept tcp on the data port (#393)
bench: emit the raw drop-identity counters to test #377's mechanism (#384)
bench: record the commit every published number was measured at (#383)
bench: add a TCP arm — saturating BulkSend, with ##GOODPUT## as its headline (#381)
bench: regression floors for the grid arms — and why they are not anchors (#380)
bench: add a Nakagami fading channel — the other half of the v1.4.0 grid (#374)
bench: add a mobility-model knob — steady-state RWP and Gauss-Markov (#373)
bench: emit ##CONFIG## so a run records the configuration it ran with (#370)
bench: emit ##HOLD## so pending-queue hold time is reachable, not just measured (#367)
bench: measure channel occupancy per run for the #308 per-hop hypothesis (#364)
bench: emit hopsCommon/hopsSurplus for the #308 phase-2 decomposition (#363)
core: adopt the thesis's pheromone exponents, betaAnts = betaData = 20 (#327)
bench: paired per-seed verdicts in sweep_summary --vs (#326)

bench: carve genuine losses out of the drop residual, not raw MAC drops (#388)
bench: the drop-identity overlap must use macDrops, not macTerminal (#385)
bench: reorder columns must be absent under TCP, not zero (#382)
bench: gauss-markov aborted on NormalPitch — a constant where ns-3 demands a normal (#375)
bench: pin manet-baselines rng streams and extend the seed-independence gate (#361)
bench: pin rng streams per seed so results do not depend on split structure (#355)