Ownership was checked ad hoc: some handlers loaded-and-compared
ownerId, some lib mutators enforced it in WHERE clauses and silently
no-op'd for non-owners, and nothing tied the two together. Two real
authorization bugs hid in the gaps: linkActivityToRoute ignored its
ownerId parameter entirely (any logged-in user could relink any
activity), and createRouteFromActivity loaded any activity without an
ownership check (a non-owner could copy a private activity's GPX into
their own route). PUT /api/v1/routes/:id also returned ok:true for
non-owners without updating anything.
ownership.server.ts is now the single enforcement point:
- loadOwnedRoute / loadOwnedActivity (non-throwing, for callers with
their own error vocabulary) and requireOwnedRoute /
requireOwnedActivity (throwing data() 404/403 for web handlers; 404
by default so guessed ids don't leak existence)
- the returned entities carry an Owned<> brand; mutators (updateRoute,
deleteRoute, deleteActivity, updateActivityVisibility,
linkActivityToRoute, createRouteFromActivity) now require an
OwnedRef, so skipping the check is a compile error
- vouchOwnership is the explicit, greppable escape hatch for the one
non-session authorization path (the Planner JWT callback)
- WHERE ownerId clauses stay in the mutators as defense in depth
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Rollup was warning on 5 modules that were both dynamically and statically
imported. With static importers in the same chunk, the dynamic forms
buy no chunking benefit — they were leftovers from earlier
cycle-avoidance workarounds that no longer apply.
Converted to static:
- @trails-cool/gpx (routes.\$id.server.ts, sync.import.\$provider.server.ts)
- logger.server (boss.server.ts — comment claimed test cycles, but tests pass)
- boss.server (activities.server.ts at two sites)
- connected-services/manager (komoot/importer.ts, wahoo/importer.ts)
- notifications.server (root.tsx)
Kept dynamic: fit-file-parser in sync.import.\$provider.server.ts — it's
heavy and only the FIT ingestion path needs it, no other static
importers exist, so the dynamic actually does chunk-split it.
Build is now warning-free.
Full repo: pnpm typecheck, pnpm lint, pnpm test, pnpm build all green.
192 tests passed (up from 181 — rate-limit test from #424 + others).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Completes the .server.ts split started in #418. Every route that mixes
a default-export component with a server-only loader/action now has a
sibling <route>.server.ts holding the data-fetching helpers; the route
.tsx is a thin delegator.
Routes converted (21):
activities._index, activities.$id, activities.new, auth.accept-terms,
auth.verify, explore, feed, notifications, routes._index, routes.$id,
routes.$id.edit, routes.new, settings, settings.account,
settings.connections.komoot, settings.profile, settings.security,
sync.import.$provider, sync.import.komoot, users.$username.followers,
users.$username.following
Pattern (same as home.tsx / users.$username.tsx / settings.connections.tsx):
- loader → `return data(await loadX(request, params?))`
- action → `return await xAction(request, params?)`
- All `getDb` / Drizzle schema / `~/lib/*.server` imports move to the
.server.ts sibling.
- `throw redirect(...)` and `throw data(...)` propagate through the
delegator unchanged.
No behavior changes — pure module-graph cleanup. Component modules no
longer transitively import the DB client; Vite's tree-shake of
server-only code is now backed by an explicit, file-local contract.
Verified:
- pnpm typecheck — green
- pnpm lint — green
- pnpm test — 181 passed, 31 integration-gated skipped
- pnpm --filter @trails-cool/journal build — succeeds
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
