{
  "id": 1858220,
  "title": "My QUIC transport had never once been executed. Here's what happened when I ran it.",
  "url": "https://urgent.news/2026/08/19/my-quic-transport-had-never-once-been-executed-heres-what-happened",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-19T03:35:06.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/copyleftdev/my-quic-transport-had-never-once-been-executed-heres-what-happened-when-i-ran-it-24ge"
  },
  "original_language": "en",
  "account": "I had written previously about SMESH, a coordination protocol that works like mycorrhizal networks connecting trees in a forest. Signals diffuse, decay on their own, and get reinforced when independently confirmed, leading to consensus without central control. This post details my attempts to determine if SMESH truly worked. The transport that had never been run before had a QUIC transport incorporated into it for several hundred lines of code. The quinn endpoint acted as both server and client, using self-signed certificates, length-prefixed bincode frames over unidirectional streams, connection pooling, and an accept loop that spawned tasks for each connection and stream. Every test passed, and the workspace remained green. Upon closer inspection, I couldn't find anyone who instantiated QuicTransport. The networking layer existed in tests, documentation, and code, yet never executed. Three bugs emerged during the integration test process. First, the system panicked during the first call to QuicTransport::new due to an issue with determining the process-level CryptoProvider from Rustls crate features. Second, dialled connections were write-only, causing the accept loop to miss incoming streams. Lastly, an unbounded allocation from an attacker-controlled length prefix allowed a process to allocate 4 GiB of memory when given a 4 GiB length prefix. These were not clever bugs but rather what happens when code meets a socket for the first time. The more significant issue was the flawed diffusion algorithm that assumed the entire graph was accessible from a single node. This led to the need for three corrections: acknowledging independent conclusions, treating relaying as an agreement rather than just a messenger, and implementing a proper merge rule for gossip.",
  "summary": "I've written before about SMESH, a coordination protocol modelled on mycorrhizal networks — the fungal web that lets trees in a forest warn each other about drought and disease with nothing in charge of the network. Signals diffuse, decay on their own, and get reinforced when independently confirmed. Consensus emerges instead of being orchestrated. That was the idea. This post is about the part…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}