Skip to content

v1.2.0

Released 2026-08-02 · GitHub release and downloads · every commit since v1.1.0

Features

  • bench: corridor-cell pheromone sampler — the #216 mechanism trace for arms that fail to shift (#288)
  • core: the proactive emission gate compares per-link — the #180 re-derivation, ADR-0018 (#284)
  • bench: the asymmetric-congestion corridor cell on the ISL torus — #216 cell 1 (#280)
  • ns3: per-interface transmit-queue congestion reader for the p2p/ISL regime + ADR-0017 (#206) (#277)
  • core: the ILinkState congestion queries name the next hop — the stamp moves to where the interface is known (#206) (#276)
  • core: virtual pheromone ages on the regular table's time-proportional clock (#262) (#275)
  • bench: adopt the dedicated diversity cell — floor-relative gate at the churn-free window (#273)
  • ns3: fast ISL failure detection via interface-down + failcell detect/reconverge split (#266)
  • skills: satellite results pipeline — validate, parse and commit isl-grid output (#265)
  • bench: split the benchmark surface into MANET and satellite views (#257)
  • core: land the proactive emission gate, default OFF, with its experiment (#252)
  • bench: add the thesis's eq 5.1 delay jitter alongside the FlowMonitor column (#247)
  • core: per-ant-type gates and directed reactive discovery (#243)
  • ns3: route next hops per interface and add the isl-grid scenario (#224)
  • bench: measure path length, used-path diversity and flow fairness (#217) (#223)
  • bench: break the anonymous packet loss down by cause (#215) (#222)
  • bench: measure packet reordering, the documented cost of multipath (#212) (#221)
  • bench: measure radio energy for all four protocols (#209) (#219)

Fixes

  • core: receptions refresh the 1-hop pheromone at the metric's unloaded 1-hop value — not the legacy constant 1.0 (#279) (#283)
  • bench: the corridor background transmits at constant duty — the OnOff default off-phase zeroed the congestion signal every other second (#282)
  • bench: leak-free energy accounting — integrate the PHY State trace, drop BasicEnergySource + WifiRadioEnergyModel (#271)
  • bench: let --energyJ=0 disable the energy model — the #256 OOM is its cancelled switchToOff events (#270)
  • bench: cap sim address space at 13 GiB so an OOM dies gracefully, not with the runner (#269)
  • bench: stream ##RSS## samples from the paper run — forensics that survive runner death (#268)
  • bench: keep OOM forensics — print and upload time -v + stderr on a failed benchmark run (#267)
  • ns2: add missing ns-default.tcl entries for four tcl-bound parameters (#261)
  • core: restore proactiveInterval to 10 s — #252 leaked the 2 s rate it claimed not to ship (#255)
  • skills: validate ##BENCH## cells' diagnostic lines, not just the table (#253)
  • bench: carry jitter_eq51_ms through the campaign CSV (#249)
  • benchmarks: catch the metric defects the campaigns exposed, and the gate starvation that hid them (#242)