A client that submits many transactions must distinguish delivery acknowledgement from evidence that a transaction executed. This report studies an implemented transport in the public `staccDOTsol/break-solana-chat` fork. Operator-configured RPC peers are admitted through genesis, health and slot checks. Reads race across peers; transaction writes rotate among peers and fail over with the same signed bytes. Overlapping sends of the same wire string share a pending operation, and an optional TPU submission callback races with RPC delivery. Confirmation polling is separate and batches signatures while retaining their order. Twelve existing tests were rerun unchanged against local stubs and passed. These establish bounded client behavior, including same-wire delivery, fast-empty-status handling, failover and poll batching. They do not establish live endpoint quality, TPU delivery, transaction finality or a throughput improvement. The historical catalogue account also names automatic gossip discovery. No implementation of that stage was recovered in the inspected source, which instead accepts configured URLs. That gap remains explicit. The contribution is a reproducible component study of delivery coordination and evidence boundaries, with human agency, access costs and accountability treated as evaluation questions rather than measured benefits. Historical account D2-40 receives partial implementation coverage; automatic gossip-discovery implementation remains unrecovered. Report text is CC BY4.0; bundled source preserves original MIT and Apache-2.0 markers. Jarett Dunn, independent researcher. Technical report, version 1.0. Preprint, not peer reviewed. Codex assisted source recovery, analysis, test orchestration and writing. Report text is CC BY 4.0; upstream software retains its licenses and attribution.
No takes yet. Share an insight, caveat, or question.
Jarett Dunn (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: