Watch

Fimeg adce4441cc tasks: log 07-21 findings (culver A/B/D/E, mic wedge, overview, keyboard); add 19-23

Records what the 2026-07-21/22 sessions actually found and shipped,
so the ledger matches reality:

- TASK-16: A, B, D landed; C (contact card) implemented, phone
  acceptance pending; new unit E (history-fill boot amnesia) done.
- TASK-08: shell-side state projection exists; add the maintenance-
  lease requirement so a deploy can never race the idle lock again
  (root cause of the 19:29 auto-lock-during-deploy crash).
- TASK-09: stale-lock incident — panel/IRQ alive, session-lock
  routing stale; capture both in the resume test.
- TASK-14: expand to the Auxo-style running-app surface and an
  unconditional Home tap; folds in the GLES OpacityMask and
  workspace-ratchet findings.
- TASK-17: pin the iOS-like keyboard target and the Hyprland
  key-safety hammer test.
- Mic trace path corrected to its new home.

New: TASK-19 (settings control center), TASK-20 (player video +
agent control), TASK-21 (reboot-to-Android confirmation), TASK-22
(settings/fullscreen tweaks), TASK-23 (culver locked-notification
nudge).

2026-07-24 10:21:44 -04:00

2.9 KiB

Raw Blame History

TASK-23 — Culver locked-notification nudge (the push analogue)

Status: designed 2026-07-22, not started. Depends on culver's activity orchestrator (landed 2026-07-22) and the shell's LockedHint write-back (landed same day, souveraine Session.qml).

Problem

Culver now stops its Matrix sliding sync while the session is locked (battery: locked = zero network/CPU — verified on device). The trade: no Matrix messages arrive while locked, so no lock-screen banners. SMS is unaffected (the modem is the platform's real push channel and telephony is exempt from set_active).

Android solves this with FCM push waking a brief sync (isSyncingNotificationEvent in Element X's SyncOrchestrator). SouveraineOS has no Google push — but it has something better positioned: the shell is always awake, owns the lock surface, and already talks to culver's notification pipe.

Design

A periodic, shell-driven sync nudge while locked:

Culver exposes a D-Bus method on the session bus: net.wiuf.Culver1.SyncBurst() — runs ONE sliding-sync round-trip (SyncService start → first room-list response → stop, bounded ~10s hard timeout) even while inactive, emitting notifications for anything inbound. Idempotent; a burst during active sync is a no-op.
The shell's IdleCoordinator schedules the nudge while locked, riding the idle stages it already manages:
- stage 0–1 (recently locked): every 2 min
- deeper stages: every 10–15 min, aligned with any existing wake (the point is coalescing with wakeups that happen anyway, not adding timers)
- suspend pending/inhibited: no nudges (system asleep = truly quiet)
Notification path is unchanged: culver's inbound fold already posts org.freedesktop.Notifications, which the shell renders as the lock card.

Security notes (same doctrine as timeline-architecture.md §activity)

SyncBurst grants nothing: it can only cause culver to do work it is already authorized to do. Any same-user process could equally start the app; the method is a battery lever, not an authority boundary.
The nudge cadence is shell policy, not culver policy — the component that owns idle/battery state owns the schedule. Culver stays policy-free.
No content crosses the nudge interface — notifications flow through the existing pipe with its existing redaction behavior on the lock card.

Non-goals

Real push (ntfy/UnifiedPush relay) — a later, separate task; this design is forward-compatible (a push arrival would call the same SyncBurst).
Per-room priority — the sliding-sync round already orders by recency.

Acceptance

Lock the phone, have someone send a Matrix message, wait ≤2 min: the lock card shows the banner while journalctl --user -u culver-app shows exactly one burst (start → response → stop), then silence resumes. Battery regression check: overnight locked drain within noise of the pre-nudge build.

2.9 KiB Raw Blame History Unescape Escape