OverviewFilesCommitsIssuesDocsReleasesPeers

hyperswarm-peer commits / 38fe11b

docs: gate status after D5/D7, and relink what the transport change touched
swarmd/README.md records that D5 is written and red on purpose, with a pointer to the three-layer trace, so nobody reads a red suite as a broken checkout. Suite re-run after the udx keepalive change: 17 offline gates, three_peer_swarm and session_channel on a live testnet, the JS vectors 65/65, all four swarmd gates, and both e2e gates on the PUBLIC DHT — the 100 MB fetch in 10.0s with 1414 bytes over the socket. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
38fe11b9164d0a806c89a0e950e8c9629b0904f9
by plankatron <fleeky@gmail.com> · 2026-09-10T22:32:51.000Z

Changed files (1) +9 -3

swarmd/README.md +9 -3
……@@ -63,7 +63,7 @@ wrong in ways the daemon could not distinguish from a bad daemon.
6363 daemon UdxTransport + queue + control plane runs
6464 channels Protomux channel per peer, SEND/on-message D1/D4 PASS
6565 content liboverlay: hash/serve/fetch/cancel D6 PASS
66 client/ swarmd::Client — the ONE wire implementation 28/28
66 client/ swarmd::Client — the ONE wire implementation 30/30
6767 tools/ swarmd-probe — built on it, in 181 lines
6868
6969 ./swarmd --socket /tmp/s.sock --storage ~/.swarmd # public DHT
……@@ -94,5 +94,11 @@ a hundred million, and the gate fails at 256 KB.
9494`--storage <dir>` turns the content plane on. Without it the daemon refuses
9595content commands rather than accepting a fetch it has nowhere to put.
9696
97See `../docs/SWARMD-DESIGN.md` §9 for the order and §8 for the remaining gates
98(D5 and D7 are still unwritten).
97D7 passes in both directions; **D5 is written and RED**, and what it found is a
98real hole rather than a gate bug — nothing detects a silently dead peer on an
99idle stream. `../docs/SWARMD-DESIGN.md` §8.1 has the trace through all three
100layers. Run it with:
101
102 python3 -B test/test_peer_down.py --bootstrap host:port
103
104See `../docs/SWARMD-DESIGN.md` §9 for the order and §8 for the gate table.