docs(website): the in-app channel as built (engagement Phase 7) #188
Reference in New Issue
Block a user
No description provided.
Delete Branch "docs/engagement-inapp-channel"
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?
The documentation §6.0b assigns engagement Phase 7. Companion to RunicGateway/website#175 and RunicGateway/runicgateway.com#24.
website/ENGAGEMENT.mdThe Phase 7 as-built:
instant(the only channel that does); the phase takes push'sdeliverand the web preferences screen, neither of which its own acceptance line mentions; the inbox takes/notificationsand the settings move under it;ctx.inbox.pushrespects a preference where one exists.body" guardrail by never letting markup into the column.inapp.eventseed named variables nothing supplies; the dedupe index is scoped narrower than the outbox's; the tickle'srefneeded an enqueue ordering to be worth anything; there was no retention policy at all; and the one only the live rig could see — staff had no reachable inbox, because/auth/me/notificationsis role-agnostic whileRequirePlayeris not.news.postis declared and nothing emits it through the engine.Two earlier passages amended where the phase made them false, both kept as history with the correction beside them:
ctx.inbox.push— present and throws until Phase 7" (filling it in is not a version bump; the signature is the one 1.7.0 declared).inappis declaredoff… so the first thing Phase 7 shipped would be a backlog" — the backlog it feared cannot happen against an empty table.§6.0b's Phase 7 row is ticked.
website/BACKEND_DESIGN.md/auth/metable, including whyunread-countis its own route (it is polled), why mark-read 404s identically for "no such row" and "somebody else's", and that no parameter names a user.user_notificationsin the table inventory: the user-scoped dedupe index and why it is narrower than the outbox's, whybodyis the text render rather than the email HTML, the block-role mapping, and the retention policy.(This branch was force-pushed once:
BACKEND_DESIGN.mdis committed with CRLF line endings, and rewriting it through a tool that normalised them turned a two-hunk edit into a 1714/1670 whole-file diff. Re-applied against the original bytes and staged withcore.autocrlf=false, so what you see is 43 added lines and nothing else.)Co-Authored-By: Claude.