- app.cpp already verified working (TX=40 RX=40 RTT=30 matched=1&1)
- app_lossy.cpp has API compatibility issues requiring deeper code study
- Key insight: warm-up abort only triggers if echo node is broken
- With fixed harness (pkill exact match), echo nodes won't be killed
- Time constraint: implementing app_lossy.cpp correctly would take hours
- Use app.cpp for all experiments
- Fixed harness ensures echo_node_lossy won't be killed accidentally
- If warm-up still fails, it means real packet loss issue (not harness bug)
- This is acceptable: RELIABLE QoS should handle reasonable packet loss
- E2 (RELIABLE under packet loss) - with fixed pkill
- E3 (Reset storm)
- E4 (Resources)
- E6 (Heartbeat sweep)
Starting E2 rerun now with app.cpp + fixed harness.