feat(modules): the three de-entanglement registries, with core as the registrant
All checks were successful
PR Checks / client-build (pull_request) Successful in 23s
PR Checks / server-tests (pull_request) Successful in 1m39s
PR Checks / bot-install (pull_request) Successful in 8m49s

Phase 2 PR 4 of docs/website/MODULE_SYSTEM.md §2.7. Adds server/src/modules/registries.js
and moves core's own notification streams, announce leg and users-detail routes
behind it, so the three seams §1.8 and §1.9 named are exercised on every boot
before any module depends on them.

Registering is validate-then-commit per registrant: the loader stages what a
module claims and the second pass commits it, so a module that throws halfway
through register() — or fails checkDeclared after it — leaves nothing behind.
That is the registry-side twin of PR 2's second-pass mount rule.

Four decisions, all the recommended option:

- announce legs became a child table. `announce_job_legs` replaces the
  towncrier_*/discord_* column groups, so the leg set is data: core registers
  `discord`, module-uo will register `towncrier`, and a module cannot ALTER a
  core table to add its own. Backfill is guarded on information_schema (a
  SELECT of a dropped column is a parse error, not a runtime one) and the
  columns go with DROP COLUMN IF EXISTS. Verified against the live dev DB:
  three legacy jobs migrated faithfully, three replays, no duplicates.
- `mapEvent` dropped from registerNotificationStreams. §1.8 already inverts the
  push path so a module owns fromShardEvent and calls core's publish() with a
  stream id it resolved; a second mapping mechanism was a leftover. The public
  safety filter, the kinds it reads and the streams it protects now live in one
  file and move together.
- core registers through the same staging area a module uses, via an explicit
  registries.registerCore() in app.js before modules.load().
- core's six /admin/users/:id/shard/* paths now go through the
  `admin.users.detail` slot, and getUser moved back to admin.controller.js.

Found on the way, and the reason two build tools changed:

- scripts/routeManifest.js could not decode a parameterised mount. Its
  unwinder expected `(?:([^\/]+?))`; express 4.22 emits `(?:\/([^/]+?))` with
  the separator inside the group. The branch had never run. It threw rather
  than guessing, which is what it is for.
- swagger-autogen cannot follow a route into an extension slot — the slot's
  router is created by declareSlot() and filled later, so there is no literal
  mount for a static parse. Regenerating deleted 407 lines and printed
  `Swagger-autogen: Success`, the spike's exact failure (MODULE_API.md §7.4).
  swagger/slotSpecs.js generates a fragment per filled slot and re-roots it at
  the prefix the router actually hangs at in the live app — read from the
  express stack via routeManifest's own mountPath, so the manifest and the spec
  cannot disagree. swagger/mergeSpec.js is the merge helper core owes for
  module fragments anyway (§6.1a), proved here against core's own slot first.

884 tests pass (856 before). routes.manifest.json is unchanged at 229 routes.
The OpenAPI spec diff is two lines of intent: the retry endpoint's summary, and
its `leg` no longer being a fixed enum.

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
2026-08-10 17:47:59 -05:00
parent bd749d4f1f
commit 6195c76d61
36 changed files with 1948 additions and 465 deletions

View File

@@ -1,87 +1,38 @@
// ── Announcement pipeline: pure logic ──────────────────────────────────────
//
// No DB, no network — just the decisions the worker and model make, kept here so
// they are unit-testable in isolation (server/test/announceJobs.test.js):
// • buildTownCrierText — turn a post into sidecar-safe town-crier lines
// • classifyTownCrier / classifyDiscord — map a dispatch result to done / retry
// / terminal, so a data problem fails fast and a transient outage retries
// No DB, no network — just the LEG-AGNOSTIC decisions the worker and model make,
// kept here so they are unit-testable in isolation (server/test/announceJobs.test.js):
// • scheduleAfter — exponential backoff schedule + the attempt cap
// • rollupStatus — derive the parent job status from the two legs
const { deriveExcerpt } = require('../../utils/sanitizeHtml')
// Sidecar town-crier caps, mirrored from the admin route validation
// (admin/uoLink.router.js: lines isArray({ max: 8 }), lines.* isLength({ max: 200 })).
// We pre-truncate to these so a published post never bounces with towncrier.error.
const MAX_LINES = 8
const MAX_LINE_LEN = 200
// • rollupStatus — derive the parent job status from its legs
// • legError — squeeze a client result into one error line
// • baseUrl / articleUrl — the public link an announcement carries
//
// What used to be here and is not any more: `buildTownCrierText`,
// `classifyTownCrier` and `classifyDiscord`. A leg's own text-building and result
// classification belong to the leg, and a leg is a registration now
// (MODULE_SYSTEM.md §1.8) — they live in utils/shardAnnounce.js and
// utils/discordAnnounce.js. This file is what every leg shares.
// Backoff between retries, indexed by attempts-so-far. Six attempts spread over
// ~a couple of hours; after the last one a leg is marked failed and surfaced in
// the post's admin panel. Shared by both legs.
// the post's admin panel. Shared by every leg.
const BACKOFF_MS = [30_000, 120_000, 600_000, 1_800_000, 3_600_000, 7_200_000]
const MAX_ATTEMPTS = BACKOFF_MS.length
// Trim to a hard length, appending an ellipsis only when something was cut.
function clamp(value, max) {
const s = String(value == null ? '' : value)
.replace(/\s+/g, ' ')
.trim()
if (s.length <= max) return s
return `${s.slice(0, max - 1).trimEnd()}…`
// The site's public base, used to build the link an announcement carries.
function baseUrl() {
return (process.env.APP_BASE_URL || 'http://localhost:5173').replace(/\/+$/, '')
}
// The public link that goes in the announcement. News has no per-post route
// (App.jsx only has the /site/news list), so we link the list — matches the
// pre-pipeline Discord announce behavior.
function articleUrl(baseUrl) {
return `${String(baseUrl || '').replace(/\/+$/, '')}/site/news`
}
// Build the town-crier lines: title, a one-line excerpt, then the URL. Each line
// is clamped to the sidecar's per-line cap and the whole thing to the line-count
// cap. Falls back to a stripped body excerpt when the post has no excerpt.
function buildTownCrierText(post, { baseUrl } = {}) {
const title = clamp(post.title, MAX_LINE_LEN)
const excerptSource = post.excerpt || deriveExcerpt(post.body, MAX_LINE_LEN) || ''
const lines = [title]
const excerpt = clamp(excerptSource, MAX_LINE_LEN)
if (excerpt) lines.push(excerpt)
const url = clamp(articleUrl(baseUrl), MAX_LINE_LEN)
if (url) lines.push(url)
return lines.filter(Boolean).slice(0, MAX_LINES)
}
// ── Result classification ──────────────────────────────────────────────────
// Both clients return { ok, status, error }. Map that to one of:
// done — delivered, mark the leg done
// retry — transient (shard restarting, bot down, network); back off + retry
// terminal — will never succeed as-is (over caps, bad auth/config); fail now
function classifyTownCrier(result) {
if (result && result.ok) return { outcome: 'done' }
const status = result ? result.status : 0
// 400 = over the line/duration caps (a data problem — do NOT retry).
// 401 = token mismatch, 409 = protocol mismatch (both config problems).
if (status === 400 || status === 401 || status === 409) {
return { outcome: 'terminal', error: legError(result) }
}
// 503 (shard not connected), 504 (shard timeout), 0 (network/timeout / not
// configured yet), and any other 5xx are transient — retry.
return { outcome: 'retry', error: legError(result) }
}
function classifyDiscord(result) {
if (result && result.ok) return { outcome: 'done' }
// The bot's /internal/announce collapses failures (503 = not connected,
// 400 = no news channel configured) without surfacing Discord's own
// retry_after, so there is no reliable terminal signal to key on here. Retry
// every failure on the shared backoff; a genuine config problem simply
// exhausts its attempts and lands as `failed` in the admin panel, where the
// per-leg retry button re-runs it after the channel is set.
return { outcome: 'retry', error: legError(result) }
function articleUrl(base) {
return `${String(base || '').replace(/\/+$/, '')}/site/news`
}
// Every leg's client returns { ok, status, data, error }. Squeeze a failure into
// the one line stored in announce_job_legs.last_error and shown in the panel.
function legError(result) {
if (!result) return 'no response'
if (result.status) {
@@ -99,29 +50,32 @@ function scheduleAfter(attempts) {
return BACKOFF_MS[Math.min(attempts - 1, BACKOFF_MS.length - 1)]
}
// Parent job status derived from the two leg statuses:
// done — both legs delivered
// failed — both legs gave up
// partial — at least one leg reached a terminal state while the other has not
// matched it (still pending/retrying, or the opposite terminal state)
// pending — neither leg is terminal yet
function rollupStatus(towncrierStatus, discordStatus) {
if (towncrierStatus === 'done' && discordStatus === 'done') return 'done'
if (towncrierStatus === 'failed' && discordStatus === 'failed') return 'failed'
// Parent job status derived from its leg statuses:
// done — every leg delivered
// failed — every leg gave up
// partial — at least one leg reached a terminal state without all of them
// agreeing (some still pending/retrying, or a mix of done and failed)
// pending — no leg is terminal yet
//
// Takes the list of leg statuses rather than two named arguments, because the leg
// set is registered rather than fixed (MODULE_SYSTEM.md §1.8). No legs at all
// rolls up `done`: with nothing registered there is nothing left to deliver, and
// leaving such jobs `pending` would pile up rows the worker never touches.
function rollupStatus(statuses) {
const list = Array.isArray(statuses) ? statuses : []
const terminal = (s) => s === 'done' || s === 'failed'
if (terminal(towncrierStatus) || terminal(discordStatus)) return 'partial'
if (list.every((s) => s === 'done')) return 'done'
if (list.every((s) => s === 'failed')) return 'failed'
if (list.some(terminal)) return 'partial'
return 'pending'
}
module.exports = {
MAX_LINES,
MAX_LINE_LEN,
MAX_ATTEMPTS,
BACKOFF_MS,
buildTownCrierText,
baseUrl,
articleUrl,
classifyTownCrier,
classifyDiscord,
legError,
scheduleAfter,
rollupStatus,
}