plan: open-backlog execution order (2026-08-16) #100
Labels
No labels
adr
android
area/calendar
area/design-system
area/i18n
area/jobs
area/offline
area/server
area/testing
bug
ci
duplicate
enhancement
help wanted
invalid
notifications
question
reliability
security
severity/low
severity/medium
tracking
web
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
eagraiclainne/app#100
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Tracking issue for the current open backlog (2026-08-16, decisions settled same day). The order below is the plan: quick notification kinds first while the PR #90 pattern is fresh, then the scheduler seam and the tickers that build on it, small Android parity and CI in between, and the repeat-rule work last because it is the largest and wants #94's move-notification hook in place.
All design decisions are made — recorded as comments on each issue. Summary: #92 and #95 closed as won't-do; #93 accepts per-step noise; unassigned jobs with deadlines broadcast to everyone (#91/#99); #99 is a distinct JOB_OVERDUE firing daily until done; #98 uses per-event lead times (plus whole-family birthday broadcasts and a day-before nudge for deadline-type events), which widens #97 to arbitrary next-fire times; #77 edits the full rule for the whole series and keeps surviving overrides.
Cluster A — notification kinds (PR #90 follow-ups)
Each kind carries the usual plumbing: proto enum, notify method, en+ga messages, webhook decorator, web push-preference checkbox + PUSH_KINDS entry, Android channel.
#92 MEMBER_JOINED— closed won't-do: joins do not need a notification.EventService/Update; fire only when the when changes.ItemService/Createwhen the parent job's owner is not the actor. Decided: accept the noise, no dedup.#95 shared-list activity— closed won't-do at the design gate.Cluster B — reminder tickers (reminders audit)
Cluster C — Android parity
deleteEventthrough the offline queue, danger action + two-step confirm, matrix-gated. No server change.Cluster D — CI
Cluster E — recurrence
Order
#94 → #93 → #97 → #91 → #99 → #98 → #77, with #78/#83/#76 slotting in anywhere. Clusters are independent stacks; within a cluster the order above is the dependency order. Bottom-first merges, fast-forward, per the usual stacked-PR flow.
Status 2026-08-16 (evening)
Everything above is built and stacked as PRs #101 → #102 → #103 → #104 (merge bottom-first; Forgejo retargets down the chain). Every item was verified against the live deploy except the Android-only UI (#78 delete flow, #83 confetti, the #98/#77 sheet controls), which awaits a connected device. PR #102 also carries an events.update offline-replay fix found during #78.
mainto v1.5.0