Closed sandreae closed 1 month ago
Attention: Patch coverage is 17.52577%
with 240 lines
in your changes missing coverage. Please review.
Project coverage is 92.46%. Comparing base (
ba416ea
) to head (98848dc
). Report is 58 commits behind head on bump-libp2p.:exclamation: Current head 98848dc differs from pull request most recent head 2798270
Please upload reports for the commit 2798270 to get more accurate results.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
(needs a rebase against main
and CHANGELOG.md
entry)
Before this PR a peer only registered on any configured relay nodes on startup in an initiation phase before the main networking event loop started. The main reason for this is that listening on the circuit relay would sometimes (quite often) fail and so we needed a retry loop which would timeout after x seconds. This pattern was clunky and not possible to fit into the event loop. It also meant that we had to disable the
Peers
behaviour, enable it, and then force connect again to the relay in order to trigger replication... ugh...Since https://github.com/p2panda/aquadoggo/pull/631 listening on the circuit relay never seems to fail, and so moving all of the relay connection and registration logic into the main event loop is now possible (yay!). This unlocks some nicer patterns around handling relays and means we can dynamically register on relays during runtime. No disabling peers behaviour and secondary force connecting required.
Relay
struct for holding current state relating to relays and a node's registration on itprintln!(..)
orinfo!(..)
startup infocheck out these beautiful logs!!
📋 Checklist
CHANGELOG.md