Participant Relationship Protocol

Back to matrix

E04-BL: Yes

Rationale

Moritz Warning publishes external meshnet-lab measurements of Babel 1.13.1, with Debian 11.7/Linux 5.10.179, Ryzen 9 Pro 3900 hardware, configuration scripts, topology files and raw CSV results. The Babel launcher assigns ULA IPv6 addresses and starts babeld on the emulated interface; the convergence test uses ICMP probes, matching the selected IPv6/ICMPv6 composition without an added security or application layer. Three published CSVs contain 30 observations each for 49-node grid and 50-node line/tree topologies. Each observation sends 200 probes after an increasing startup delay. At the final 58-second delay all three record 200 replies; these are publisher measurements, not editorial reruns. The experiment comes from a separately maintained comparative laboratory, not the Babel implementation release team. The published method and artifacts support reproduction of this bounded convergence/packet-delivery measurement.

Assessment method

Independent report, pinned scripts and raw CSV review; source-pointer reconciliation, no new execution

Source context

Moritz Warning, meshnet-lab 2024 comparative measurements

Outstanding qualification

The results README links test revision e52fbb6, whose convergence driver selects only Yggdrasil versions. The recovered general driver at 3f53e44 differs solely in that protocol-list loop and includes Babel; the tested-version declaration, retained Babel launcher and CSV names identify the Babel experiment. Reproduction requires selecting Babel in that loop and installing declared Babel 1.13.1, not blindly running the Yggdrasil-only setup. Retain this source-pointer discrepancy. The driver uses a 100-Mbit/s TBF with latency parameter 1ms; that parameter is a queue bound, not proof of a 1ms propagation-delay model. Namespace emulation, host load, topology, sampling and configuration limit transfer to physical networks. No throughput ranking, global-scale result, G01 hierarchy measurement, present-version performance guarantee or exact replay of numeric outputs is claimed. Independent attribution rests on the external laboratory and identified maintainers; it is not a conflict-of-interest certification. No local benchmark was run.

Public sources

Editorial evidence references

The following identifiers refer to research records not included in this public package; they are not public links.