Wahoo's POST /v1/routes expects route[file] as
'data:application/vnd.fit;base64,<base64>'. We were sending raw
base64, which Wahoo silently discarded — the route would appear in
the user's list with metadata (distance, ascent come from request
fields) but file.url stayed null and the Wahoo app rendered no track.
Confirmed by re-fetching route 50197876 via GET /v1/routes/:id.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The Course message advertised capabilities=0x04 (time only), telling
consumers the course had no position data. Wahoo accepted the route
(distance/elevation come from request fields) but rendered an empty
map. Set capabilities to valid|distance|position (0x1A).
Also write a real lap totalElapsedTime/totalTimerTime derived from the
1Hz record timestamps; a zero-duration lap can be rejected as malformed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
GET /v1/routes/:id needs routes_read; routes_write only covers writes.
Existing connections will need to reauthorize to gain the new scope.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Disconnecting a sync provider now calls the provider's revoke endpoint
(DELETE /v1/permissions for Wahoo) before dropping the local row, so
tokens don't accumulate against Wahoo's per-(app,user) cap. The token
exchange now classifies Wahoo's "Too many unrevoked access tokens" 400
as a distinct OAuthError code, and the connections settings page shows
a localized banner for that and other connect failures instead of a
silent ?error=sync_failed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wahoo's /oauth/token endpoint returns 400 for JSON bodies. OAuth 2.0
requires application/x-www-form-urlencoded for token requests; switch
exchangeCode and refreshToken to URLSearchParams. Also include the
response body in the thrown error so future failures are diagnosable
without scraping container logs.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The connect loader was setting state to a raw user.id, but the callback
parses state as base64-JSON via decodeOAuthState and falls back to {}
when the parse fails. That sent users back to /settings, which now
redirects to /settings/profile, hiding the freshly saved connection on
a different tab.
Encode the state as a proper PushOAuthState with returnTo set to
/settings/connections so the callback lands on the page that initiated
the flow.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The deps stage was missing the COPY for packages/fit/package.json, so
@trails-cool/fit was absent from the workspace at install time and
@garmin/fitsdk could not be resolved during the journal build.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Move completed wahoo-route-push change to archive and sync delta specs:
update wahoo-import for routes_write scope, add new wahoo-route-push capability.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Route detail page now renders one of three states for the owner of a
route with a Wahoo connection:
- "Send to Wahoo" button + privacy tooltip when no successful push
exists for the current version, plus the inline last-error blurb
if the previous attempt failed
- "Sent to Wahoo on <date>" pill when sync_pushes has a pushedAt
for the current version
- Banner row at the top mapping the ?push= and ?code= query params
the action route appends to user-facing copy
i18n strings cover all 8 banner states (success, needs_permission,
no_connection, no_geometry, validation, rate_limit, token_expired,
generic) in English and German.
Privacy disclosure lives at /legal/privacy (apps/journal/app/routes/
legal.privacy.tsx), the in-app surface this repo uses instead of a
docs/privacy.md manifest. The new bullet declares that route
geometry, name, and description are transmitted to Wahoo on opt-in
via the Send to Wahoo button.
E2E tests for the push and re-auth flows (tasks 9.4/9.5) are
deferred — they need a server-side Wahoo mock harness that doesn't
exist yet. Slice-4 unit tests cover the pipeline against the
provider mock.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds /api/sync/push/:provider/:routeId as the user-triggered entry
point for pushing a route. The route action delegates to a new
pushes.server.ts pipeline that:
- resolves the latest route_versions row and uses that GPX (not
routes.gpx) so the bytes Wahoo gets match the snapshot the user
sees
- short-circuits on (user, route, version, provider) idempotency:
a successful prior push returns the existing remote_id without
re-calling Wahoo
- detects scope_missing before hitting Wahoo and redirects through
getAuthUrl with a base64url-encoded state carrying pushAfter +
returnTo
- refreshes tokens once on PushError({ code: "token_expired" }) and
retries the push, then updates sync_connections in place
- records every outcome in sync_pushes (insert on first attempt,
update on retry) so the UI can show success/failure state
The OAuth callback handler now decodes the state, resumes a
pushAfter pipeline server-side after exchangeCode, and handles the
?error=access_denied path with a needs_permission notice.
Also flips packages/fit/src/fitsdk.d.ts to a regular .ts side-effect
shim so journal's tsc picks up the @garmin/fitsdk module declaration
when consuming the workspace package via source.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Extends the SyncProvider interface with an optional pushRoute method
plus PushRoutePayload, PushRouteResult, and a typed PushError so the
slice-4 action route can map error codes to user-facing copy without
parsing HTTP statuses itself.
Wahoo provider:
- adds routes_write to the requested OAuth scope set
- implements pushRoute: base64-encodes the FIT, builds the form body
Wahoo's POST /v1/routes expects, parses the remote id out of the
response, and throws typed PushError on 401/403/422/429/5xx
- saveConnection now persists the requested scopes as grantedScopes
on exchangeCode (Wahoo doesn't return a scope field, so the
requested set is the granted set)
Token refresh on 401 is deferred to the slice-4 action route — same
pattern as the existing webhook handler in this repo, which does
inline refresh rather than wrapping every call.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
sync_pushes tracks outbound route pushes to providers, keyed by
(user_id, route_id, route_version, provider). Successful rows hold
the remote_id; failed rows hold the error and can be retried in
place. The unique index makes idempotent push trivial.
granted_scopes records the OAuth scope set we requested at
exchangeCode time. Wahoo doesn't return a scope field in token
responses and grants scopes all-or-nothing, so the requested set is
the source of truth. Defaults to an empty array, which means any
pre-existing connection will be flagged as scope-mismatched on the
first routes_write push — the intended UX for the slice 3 re-auth
flow.
Adjusts tasks 2.3/2.4 in the spec to match repo reality: this
project runs `drizzle-kit push --force` schema-first, with no
checked-in migrations directory.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wraps @garmin/fitsdk to emit FIT Course files from GPX, the binary
format Wahoo's POST /v1/routes API requires. Server-side only — the
~1 MB SDK never ships to the planner browser bundle.
Round-trip tests use fit-file-parser as an independent oracle and
assert lat/lon parity within 1e-4 deg and altitude within 0.5 m
across short flat, alpine, multi-day, and single-point fixtures.
Updates design.md decision #1: the original hand-rolled-encoder plan
was justified largely by ESM friction in the Garmin SDK, but as of
v21.202.0 the SDK is pure ESM with zero deps. Wrapping it saves us
~400 LOC of binary plumbing and ongoing maintenance against future
FIT spec updates.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Update sync_pushes schema to use text ids matching the existing journal
tables, and fix the schema path references to packages/db/src/schema/journal.ts.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Wahoo's Cloud API exposes POST /v1/routes (scope routes_write), which
syncs to the Wahoo App and to ELEMNT/BOLT/ROAM head units. This change
proposes the push pipeline: GPX -> FIT Course (new @trails-cool/fit
package) -> Wahoo, with sync_pushes for idempotency, an explicit
re-auth flow for the new scope, and a "Send to Wahoo" affordance on
the route detail page. Read-only Wahoo import is unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Roughly tripled the demo persona's name + description pools so the
demo bot's daily output doesn't feel canned within a few weeks of
watching it. Same Bruno: dog-park-inspector deadpan, Berlin-set,
absurd bureaucratic tone. New entries grouped into clear categories
inside the file (more Berlin neighborhoods, time-of-day variants,
bureaucratic-deadpan one-liners, specific events, wildlife
encounters, stick + tennis-ball lore, self-aware moments, Berlin
weather) so it's easy to add to one bucket without rewriting the
whole list.
Counts before → after:
- en names: 12 → 35
- de names: 12 → 35
- en descriptions: 10 → 30
- de descriptions: 10 → 30
No schema or behavior change. Tests unchanged — the persona schema
still parses; the existing pool-shape-agnostic tests still pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Outcome of today's `/spec-drift-review`. Concentrated on the
high-and-medium-severity items; low-severity wording left for the
next per-feature change to pick up naturally.
High-severity URL fixes (specs were actively misleading):
- account-management — email-change verification URL was specced as
`/auth/verify-email-change?token=...`; the actual route is
`/auth/verify?email-change=1&token=...` (see auth.verify.tsx:8).
A reader implementing against the old wording would build a link
that 404s.
- observability — "Both apps SHALL expose a /metrics endpoint" was
half right: the planner is at /metrics, but the journal exposes
/api/metrics. The infrastructure spec already had the right URLs;
the observability spec disagreed with itself. Now both correct,
with a one-line note explaining the per-app split.
Medium-severity wording drift (Stream E aftermath):
- profile-settings, account-management, connected-services — all
three said "the settings page SHALL include a [...] section",
which described the old single-scrollable settings page. Stream E
(PR #323) split /settings into four sub-pages
(/settings/{profile,account,security,connections}); rewording each
spec to point at its specific sub-page. The API endpoints
(/api/settings/*) and behavior are unchanged.
- authentication-methods — passkey add/delete previously said "via
the settings page"; now specifically /settings/security.
Code drift fixed inline:
- apps/journal/app/routes/auth.verify.tsx — after a successful
email change, the redirect was going to `/settings#account` (an
anchor on the OLD single-scrollable settings page). Stream E
retired that page; the right destination now is
`/settings/account`. Without this the user would land on
/settings/profile (which is what /settings redirects to) instead
of the page that just changed.
sse-broker spec wording:
- The "no buffering tweaks in the Caddy reverse_proxy block"
scenario asserted the entry was "a plain `reverse_proxy
journal:3000`". After PR #329 the journal block has
`lb_try_duration 30s` / `lb_try_interval 250ms` — neither affects
streaming, so SSE still works, but the spec's "plain" language
was no longer literally true. Reworded to forbid only
buffering-related directives; explicitly call out that
retry-on-restart directives like lb_try_duration are fine.
Navbar consolidation (journal-landing):
- The shipped navbar's full shape was scattered: notifications said
"navbar has a bell," explore said "navbar has an Explore entry,"
but no spec described the avatar dropdown, the primary-nav
cluster (Feed/Routes/Activities), or the mobile drawer (Stream
C / PR #324). Added a "Top navbar shape" requirement to
journal-landing covering all of it — anonymous vs signed-in,
desktop vs mobile, dropdown contents, drawer behavior. The
per-feature specs (notifications, explore) still own their own
badges/entries; this requirement just says what the whole
cluster looks like.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
cd-apps already scp's `infrastructure/Caddyfile` to the server
alongside docker-compose.yml — but the running Caddy container
doesn't auto-pick-up config changes. cd-infra is the workflow that
calls `caddy reload`, and it only triggers on `infrastructure/`
paths. So any Caddyfile edit shipped through cd-apps (e.g. the
`lb_try_duration` block in PR #329, deployed as part of an apps
push) sits on disk unapplied until the next cd-infra run.
This was real today: PR #329 added `lb_try_duration 30s` to silence
deploy-time 502s. PR #331 (a vite.config edit, apps path) merged
right after and triggered cd-apps, which copied the new Caddyfile
but didn't reload Caddy. The very next planner restart in that
deploy promptly produced three 502s — the exact thing #329 was
supposed to prevent. We applied the reload manually via SSH after
the fact.
This commit makes that reload happen on every cd-apps deploy. The
operation is idempotent (Caddy validates first, swaps live, no
downtime) so doing it even when Caddyfile is unchanged costs
nothing. `|| true` keeps the deploy from failing if Caddy is
unhealthy at deploy time.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds `server.warmup.clientFiles` so Vite eagerly transforms the
journal entry + every page route at server startup. Without it,
the first visit to a previously-unseen route triggers Vite's
"new dependencies optimized → reloading" path, which is a full
page reload — fine as a one-time blip in everyday dev, painful
when an e2e run is hitting many routes back-to-back on a fresh
server (each new route reload races with whatever the test was
doing).
This is a partial improvement, not a complete fix, for the local
auth e2e flakiness investigated earlier today. The deeper issue
there is a separate React-hydration-vs-Playwright-click race in
dev mode (production CI doesn't see it because production JS
hydrates faster). That needs a different fix at the test/fixture
layer; this commit just removes one source of noise from the
investigation by eliminating the dep-discovery reload.
Glob picks up `.tsx` files only — `.ts` route handlers under
app/routes/ are server-only API endpoints and Vite refuses to
transform them for the client (errors "Server-only module
referenced by client"). Page components alone drive the
dep-discovery we care about.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Most contributors don't need HTTPS=1 locally — plain HTTP is the
default and the right choice for everything except Wahoo OAuth
callback testing. WebAuthn (passkeys), magic links, sessions, the
Terms gate, SSE all work over HTTP because the WebAuthn spec treats
localhost as a secure context regardless of scheme. CI proves the
point: the e2e suite runs over plain HTTP and passes cleanly.
The original HTTPS=1 plumbing landed in 20b91ef (2026-04-05) bundled
into a Wahoo import fix, with no inline rationale. Once you've
forgotten the reason it tends to leak into the default workflow,
which then breaks the local e2e suite (Playwright always uses HTTP
baseURL; an https ORIGIN env mismatches what it sends) and creates
unnecessary divergence from CI.
Documenting the single legitimate use case so the next contributor
(or future-me) doesn't have to re-derive it from git blame:
- apps/journal/vite.config.ts — expanded the comment near the
basic-ssl plugin to spell out:
* What works on HTTP (everything except Wahoo)
* What needs HTTPS=1 (Wahoo OAuth specifically)
* The norm: don't add new HTTPS-only paths without writing them
down here too, so the assumption stays auditable
- apps/journal/.env.example — new tracked file showing the env vars
the journal app reads, with `ORIGIN=https://localhost:3000`
marked as HTTPS-only and accompanied by an explanation of the
ORIGIN/Playwright mismatch trap. Future contributors don't have
to discover this through a failing e2e run.
- CLAUDE.md — new "Local HTTPS dev (rare)" subsection under
Development Commands. Includes the exact command for Wahoo testing
(`HTTPS=1 ORIGIN=https://localhost:3000 pnpm --filter
@trails-cool/journal dev`), the turbo-doesn't-pass-HTTPS gotcha,
and the rule of thumb: don't set ORIGIN in your .env unless you
also always run with HTTPS=1.
No code/behavior changes; documentation only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The token lives in `secrets.infra.env`, which `cd-infra.yml` merges
together with `secrets.app.env` into the server's
`/opt/trails-cool/.env`. The cd-apps workflow's own `app.env`
intentionally does NOT carry the token (apps don't need it at
runtime), so the original `grep ... .env` was correct. My earlier
edit in this branch swapped the path to `app.env` and would have
broken the annotation hook the moment it actually worked.
Restored `.env` and updated the inline comment to make the file
ownership explicit (cd-infra populates it; cd-apps reads it).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The journal/planner deploy in cd-apps.yml does `docker compose up -d
journal planner`, which stops the old container and starts the new
one — Caddy keeps forwarding requests during the ~10–30s gap and
returns 502s. The caddy-502-rate alert (threshold > 0 for 2m)
correctly trips, every time.
Two production changes plus a long-broken workflow detail:
- infrastructure/Caddyfile — add `lb_try_duration 30s` /
`lb_try_interval 250ms` to the journal and planner reverse_proxy
blocks. Caddy now holds and retries the upstream for up to 30s
during a restart instead of 502'ing immediately. Real outages
(upstream unreachable longer than 30s) still 502 and the alert
still fires for those.
- infrastructure/grafana/provisioning/alerting/alerts.yml — add a
comment documenting why caddy-502-rate stays at threshold > 0:
with lb_try_duration in front of it, the alert no longer
conflates "deploy in flight" with "real outage."
- .github/workflows/cd-apps.yml — fix a long-silent bug: the
Grafana deploy-annotation step was reading
GRAFANA_SERVICE_TOKEN from `.env`, but the secrets file we scp
to /opt/trails-cool is named `app.env`. The token check failed
silently and the curl was being skipped on every deploy.
Switching to `app.env` so deploys actually annotate Grafana.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The directory was filtering out the demo persona on the rationale
that "the demo bot is not a real user and should not appear in real
discovery." That's exactly backwards — the whole point of having a
demo persona is to give new users a follow target so the platform
doesn't feel empty when they arrive. Hiding the bot from the
discovery surface defeats its purpose.
Concretely on flagship: only one local user (ullrich) was visible
on /explore today, even though Bruno (the demo persona) is
public-by-default and posting public activities. After this change
both appear; Bruno carries a small "🐕 Demo account" badge next to
his display name so viewers know what they're following.
- apps/journal/app/lib/explore.server.ts — drop the
ne(users.username, persona.username) clause from exclusionFilters.
The demo persona is now treated like any other public user. Banned/
suspended scaffolding stays for forward-compat.
- apps/journal/app/routes/explore.tsx — loader computes isDemoUser
per row (cheap, just username comparison against
loadPersona().username). DirectoryRow renders the demo badge inline
with the display name, matching the existing pattern on
/users/:username.
- openspec/specs/explore/spec.md — updated the "Excluded users"
requirement to remove the demo persona, replaced the "demo
excluded" scenario with "demo appears with badge", and updated
the "Active recently" requirement + scenarios accordingly.
- apps/journal/app/lib/explore.integration.test.ts — flipped the
demo-persona test from "is excluded" to "is included".
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
