Audit connection reliability and reconnection #60
erik-sutton95
started this conversation in
Ideas & Feature Requests
Replies: 1 comment
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Ongoing audit of the BLE → SoftAP → UDP spine: stay connected, recover cheaply, never black the feed with a bad repair.
This is audit first, then small remaining PRs. Most of
docs/connection-reliability-plan.mdis already in tree. Do not re-implement the types.Related: Android parity (workstream 7) · Multicast · roadmap index
Already implemented
LinkDiagnosis,ConnectTimeline,FeedWatchdog,SessionRecovery,ReconnectBackoff,CameraLinkHealth, SoftAP persist helpers0x09/0xa8)failLink→ FAILEDHard rules stay: do not spam
0x09/0xa8; do not tear UDP because a SET timed out while HEVC/AVC is still arriving.Audit (no behavior change)
LinkDiagnoservsFeedWatchdog.tickvsCameraSessionbranches.LinkDiagnoseris unused outside tests;.rehandshakenever maps fromrepair(for:)rebuildVTSessionin the watchdog currently rebuilds UDP, not VTfeed: stall/feed: black) on Pocket 4 Pro and Nano, LUT on, AF-C hunt, background/foregroundgatt/pair/path/hs/enable— isidr=still worth adding?Remaining implementation (after the audit, one PR each)
LinkDiagnoseror delete/narrow it)DatalinkDriver.openleftover)FeedWatchdog+SessionRecoveryover JNIdocs/feed-watchdog.mdif the 3–5 min black well is gone or still openDone when
A written audit comment on this thread, plus the leftover PRs filed or closed. Hardware proof that a 5+ min take with LUT survives congestion without a GOP reset.
All reactions