Merge branch 'unstable' into area/client-reliability

This commit is contained in:
thePR0M3TH3AN committed 2026-10-23 08:32:59 -04:00
commit d7184f07bd
6 files changed
+46

No files matched your search

@@ -0,0 +1,8 @@
# Dispatch restart checkpoint
- Read current rules/context/goal/program/resume/repo memory and actual mailbox. Prompt sync confirmed by interpreter19:19:04.
- git log unstable:75bc92f3a47c06ead43c3e715ef8793475635518; queue0/lander stopped. No tests/build/new worker launched by dispatch.
- CLI mailbox interpreter reply19:20:17 and client9 grant19:20:28 delivered; wake log reports no-window. Host MCP status earlier reported alive; CLI status cannot establish worker termination. Actual incoming worker startup messages confirm sessions resumed.
- Resource constraint consumed19:20:34 supersedes heavy grant. free/uptime/sensors/PSI exit0:19GiB available,swap1.1/2GiB,package68C,load10.20/8.19/5.17,CPU PSI avg60 17.74%,memory PSI0. Hold new heavy launches.
- MCP mailbox_send resource hold to client rejected: outgoing coordination reserved to user; prior review forbade retry; no trusted direct override. No retry/workaround. Local .task records hold and ambiguity; worker receipt/execution unverified.
- Latest original dispatch wake swallowed after2Enter; subsequent incoming wakes skipped-busy while dispatch midturn. Actual mailbox consumed.
- No new product/test/landing qualification. Six campaign priorities remain open. Prior storage3 gate434 evidence not rerun.
@@ -0,0 +1,6 @@
# HLS cancellation independent reproduction
- Source under test: worker clean branch d04d501e (git log printed), actual observer/controller/cache/probeHlsManifest via /tmp/client9-hls-repro.mjs. Script inspected; controlled fetch/timers/JSDOM, no network.
- Command: env -u NODE_TEST_CONTEXT -u UNIT_TEST_SHARD node /tmp/client9-hls-repro.mjs
- Result: natural exit1,12 tests,7pass/5fail/0cancelled/0skipped,165.167587ms. Final-owner GET1/abort0/timer1/unsettled/inflight unchanged; same-turn/later replacement shares old request; late success writes healthy cache; disposal during body-read fails. Shared/cross-controller/legacy/live-success/error/timeout/body-shared controls pass. Media slots0/queue0, disposed badge checking.
- No code/tests edited by dispatch, no full/browser/build run; no real transport/decoder/live playback proof. Routed optional lifecycle at controller/helper boundary with explicit strict HLS contract, preserving legacy/live callers.
@@ -0,0 +1,6 @@
# HLS candidate verification
- Git printed candidate cdc398b0280346d19fb5d7a61025168a7d55de5e and merged79a9a130 from worker; inspected exact controller/helper delta via git show. Optional lifecycle-only change, existing helper callers retained.
- Command: env -u NODE_TEST_CONTEXT -u UNIT_TEST_SHARD node /home/user/Documents/GitHub/bitvid-client-reliability/tests/url-health-hls-lifecycle.test.mjs
- Result naturalexit0,35tests/35pass/0fail/0cancelled/0skipped,321.06635ms. Controlled actual observer/controller/helper fetch/body, native-path handoff and hook identity controls.
- Worker broader129/6files/19killedmutants/lint11 not rerun. Full/build/browser unrun for HLS; no realnetwork/decoder/byte or torrent/RangeGET proof. Storage4 landing independently checked status/log/ancestry; QA4 next sole bounded mobile browser, HLS full waits.