Refactor the CI pipeline to use WithServiceBinding for the Stalwart mail
server, replacing legacy shell scripts and manual port management.
Introduces pre-seeded data for the Stalwart service to avoid network
hits and improves headless UI testing with Xvfb.
The CI self-hosted runner can leave a stalwart process alive from a prior
run that was interrupted externally, causing the next run to fail with
"port already in use". Kill any existing stalwart before starting a new one.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The builds page at /builds/ was empty because generate-build-history
only ran inside deploy-playstore; if that job failed early (e.g. Play
Store secrets not configured) the website was never updated, and the
build-linux job never triggered a website update at all.
Changes:
- generate_build_history.py: extend to cover Linux tarballs in addition
to Android APKs, capped at MAX_BUILDS_PER_PLATFORM (30) each
- Taskfile: add website-publish task (generate-build-history +
website-deploy), exclude *.tar.gz from rsync, update descriptions
- .forgejo/workflows/ci.yml: add publish-website job that waits for
both build-linux and deploy-playstore (using always() so it runs
even when deploy-playstore fails), then removes the duplicate
generate/deploy steps from deploy-playstore
- .github/workflows/ci.yml: add deploy job that deploys Linux build,
generates build history, builds Hugo site, and rsyncs to server
- .gitignore: ignore website/content/builds/_index.md (generated),
Python __pycache__, and widget test failure screenshots
- stalwart-dev/integration_ui_test.sh: use ${USER:-$(id -un)} for
robustness in environments where USER is unset
- scripts/test_generate_build_history.py: unit tests for parse_builds
and render_entries covering both platforms
Generated content (builds/_index.md and per-day pages) is not tracked
in git; it is produced at CI time and rsynced to the server.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
With set -Eeuo pipefail, a failing fvm flutter test exited the script
before _e2e_exit=$? could run, so the retry-on-new-display logic never
fired. Use the cmd || var=$? pattern to capture the exit code safely,
and add || true to the break guard so set -e doesn't trip on it.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add --no-warn-dirty to all nix develop calls to suppress Git dirty-tree warnings
- Switch integration test reporter from expanded to compact (per-test names suppressed on success)
- Show only summary line on integration test success, matching unit/widget test behavior
Closes#8
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The U7 onboarding view replaced "No accounts yet." with "Welcome to
SharedInbox", causing the E2E test to spin for the full timeout budget
(pumping slowly in headless CI) before failing. Fix the finder and
bump per-attempt timeout from 240s → 360s and CI job ceiling from
20 min → 30 min to give the full account-add → send → verify flow
room to complete.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
xvfb-run catches SIGTERM from `timeout`, kills its children, and exits 0,
making a timed-out test indistinguishable from a pass (CI #168 false positive).
Running Xvfb ourselves captures fvm flutter test's real exit code so timeouts
(exit 124) are correctly treated as failures.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previous failed CI runs leave orphan sharedinbox/flutter processes that hold
onto Xvfb display resources, causing the next run's GTK app to hang during
initialisation (never connects back to the flutter test runner, no output
for 9+ min until timeout fires).
Fix:
- Kill stale sharedinbox/flutter processes before launching xvfb-run
- Retry the xvfb-run call once (4-min timeout per attempt) so a transient
display-init hang doesn't permanently fail the job
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Sequential CI steps leave the runner under heavier load than the parallel
task check approach, so the E2E test can legitimately take 4-5 min.
Raise timeout 300→600 in integration_ui_test.sh and step timeout 6→12 min.
Job-level ceiling raised to 30 min to match.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Split single 'Run Full Check Suite' step into named steps so per-step
timing is visible in the CI UI
- Add timeout-minutes: 20 to the overall job and timeout-minutes: 6 to
the UI E2E step — previously a stuck xvfb-run could hang for 23+ min
- Add 'timeout 300' to xvfb-run in integration_ui_test.sh so the E2E
test exits with a clear error instead of hanging indefinitely
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace fixed ports with dynamic allocation (port 0) for all Stalwart listeners, including ManageSieve.
- Require KVM acceleration for Android integration tests; fail early with setup instructions if /dev/kvm is inaccessible.
- Require all ANDROID_APK_SCP environment variables for deployment; fail early if any are missing.
- Revert emulator boot timeouts to standard values (120s device / 60s boot) now that software emulation is disabled.
- Run integration tests sequentially in Taskfile.yml to avoid port 4190 (ManageSieve) conflict.
- Add ManageSieve listener to Stalwart config for better test coverage.
- Increase Android emulator boot timeout and add software-mode flags (-accel off, -no-boot-anim, -gpu swiftshader_indirect) to accommodate environments without KVM.
- Update LATER.md with notes on software emulation performance.
xvfb-run --auto-servernum picks a fresh display number, but if a previous
session left an orphan Xvfb process (e.g. killed mid-flight), the stale
/tmp/.X11-unix/X<N> socket and X<N>-lock still belong to that orphan, and
xvfb-run's cleanup at the end fails with "kill: No such process" — flipping
the script's exit status to non-zero even when the integration test itself
passed. Reap orphan Xvfb processes (pgrep -u $USER -x Xvfb) and remove their
display sockets and lock files before invoking xvfb-run.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The Android UI integration test failed at tap(aliceTile) with "0 widgets"
even though pumpUntil had just found the tile. On the slow software-rendered
emulator the route-pop animation finalises during pumpUntil's trailing 300 ms
settle, briefly leaving the tile out of the tree. Re-confirm with a second
pumpUntil before the tap.
Bundles the previously uncommitted infra changes that make task deploy-android
run end-to-end inside nix develop: Linux desktop runtime libs + GL software
rendering env in flake.nix, path_provider_android pin to <2.3 to avoid the
libdartjni SIGSEGV, deferred DB-path resolution after WidgetsFlutterBinding,
+iglx for xvfb-run, platform-tools on PATH, and a single pre-commit script
replacing the dart-format / task-check-fast pair.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
On Android, the soft keyboard keeps viewInsets.bottom non-zero while the
search TextField is focused. ListView.builder is allocated near-zero
height and renders 0 items, so find.text(subject) always finds nothing
even though the IMAP search returned results. Unfocusing the primary
focus after enterText dismisses the keyboard and gives the results list
full body height before pumpUntil starts polling.
Also fix pumpUntil to use pump(300ms) instead of pumpAndSettle() so a
continuously-running animation (spinner under CPU load) never prevents
settling, and override accountConnectionStatusProvider so _AccountTile
never shows a CircularProgressIndicator during the test.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
CircularProgressIndicator in _AccountTile (from accountConnectionStatusProvider)
runs continuously and prevents pumpAndSettle() from ever settling on Android,
causing frame-pump storms that drop the StreamBuilder data state and make
tap(aliceTile) find 0 widgets.
Overriding the provider to return immediately means no spinner ever enters the
tree, so pumpUntil() can use pumpAndSettle() cleanly again.
Also adds task run-android (boots sharedinbox_test AVD and runs flutter run).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- pumpUntil uses ListTile-scoped finder so it doesn't exit early when
'Alice' is still in the form's EditableText before navigation pops
- tap(aliceTile) reuses that same finder instead of a second find.text
- EmailListScreen search bar adds onChanged debounce (300ms) so the
test never needs receiveAction(TextInputAction.search), which caused
a keyboard-dismiss animation that triggered layout overflow in
disposed render objects
- FlutterError.onError filter in the test suppresses DEFUNCT/DISPOSED
overflow errors from Android's route-teardown layout passes
- integration_android_test.sh: force-stop + pm clear before uninstall
so stale app data can't bleed into subsequent runs
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add INTERNET permission to main AndroidManifest.xml (was missing from
release builds, causing all network calls to fail on device)
- Add scripts/mobsf_scan.sh: uploads release APK to MobSF after each
build and asserts required permissions are declared; docker pull -q
suppresses progress-bar noise
- Wire MobSF scan into build-android task; add mobsf-stop convenience task
- Fix _AccountTile subtitle overflow on Android: replace Column([Text,Text])
with single Text('email\ntype') so ListTile can measure height correctly
- E2E test robustness on Android: use pumpUntil(find.text('Alice')) instead
of pumpUntil(FAB)+expect to handle Drift background-isolate stream delay;
add skipOffstage:false to tap; remove stale email-address assertion
- Uninstall app before each Android integration test run to clear leftover
DB state and prevent "Unable to start the app" on repeated runs
- Update widget tests to use find.textContaining for merged subtitle text
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Platform.environment is empty inside the Android app process, so the dynamic
IMAP/SMTP port numbers exported by the test script were never visible to the
Dart test code. The test fell back to its defaults (127.0.0.1:1430/1025),
which aren't reachable inside the emulator.
Replace the export STALWART_IMAP_HOST=10.0.2.2 approach with
adb reverse tcp:1430 tcp:$STALWART_IMAP_PORT
adb reverse tcp:1025 tcp:$STALWART_SMTP_PORT
so the emulator's loopback ports 1430/1025 forward to the actual random host
ports — matching the test's hard-coded defaults exactly. Clean up the
forwarding rules in the EXIT trap.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Stalwart 0.14.x does not increment HIGHESTMODSEQ when new mail arrives
via SMTP delivery, so the incremental sync's CONDSTORE fast-path saw
serverModSeq == storedModSeq and returned early — silently skipping
UID SEARCH and missing any newly received messages.
Fix: remove the early-return fast-path. Incremental sync now always
runs UID SEARCH UID ${lastUid+1}:* to discover new messages. CONDSTORE
is still used for the flag-refresh gate (only runs when modseq changed),
which is its correct, narrower role.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds stalwart-dev/integration_android_test.sh which starts Stalwart on
random ports, detects a connected Android emulator via adb, sets
STALWART_IMAP_HOST=10.0.2.2 (emulator-to-host alias), and runs the
existing integration_test/ suite on the emulator.
Wires it up as `task integration-android` in Taskfile.yml.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add lcov to nix flake (required for flutter --merge-coverage)
- stalwart-dev/test.sh: collect and merge coverage when unit baseline exists
- run_unit_tests.sh: remove inline coverage check (now in dedicated task)
- Taskfile: add coverage task; check runs test → integration → coverage
sequentially so the gate sees combined unit + integration data
- check-fast (pre-commit) omits coverage gate since integration tests
don't run there; full gate runs only in task check
- Drop two untestable fake-only tests (UID-validity reset, malformed envelope)
- Coverage threshold restored to 80% (84% with merged data)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Root cause: flutter test ran all 3 integration test files in parallel
against the same Stalwart instance. Concurrent SMTP/IMAP from
email_repository_imap_test and concurrent_sync_test caused SMTP rate
limiting (4th send hung for ~27s) and flushPendingChanges race failures.
Fixes:
- stalwart-dev/test.sh: add --concurrency=1 so test files run serially
- concurrent_sync_test: reduce timeout 2 min → 30 s (tests now pass in ~2s)
- imap_client_factory + test helpers: set defaultResponseTimeout=20s on
ImapClient so individual IMAP commands never block indefinitely
- jmap_client: reduce HTTP call timeout 30 s → 10 s (local server; keeps
stacked-timeout total well below any reasonable per-test limit)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Fix HOME override that caused FVM to re-download 220MB Flutter SDK on
every run; use XDG_DATA_HOME instead to isolate app data without
touching HOME
- Switch DB path from getApplicationDocumentsDirectory() to
getApplicationSupportDirectory() so XDG_DATA_HOME isolation works and
stale accounts don't leak between test runs
- Replace fixed pump(5s/3s) waits with pumpUntil() polling at 200ms so
tests stop waiting as soon as the UI is ready (23s of dead wait → 8s)
- Add timing instrumentation (ts() in shell, _log()/Stopwatch in Dart)
- Fix CI integration-ui job: was mixing subosito flutter with fvm flutter;
now uses fvm consistently with actions/cache for ~/.fvm, ~/.pub-cache,
and build/linux
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- flake.nix: Flutter 3.41.6, Android SDK, Stalwart, GTK3/build
tools for Linux desktop, go-task
- .envrc: copied from sharedinbox — use flake + dotenv_if_exists
- Taskfile.yml: analyze, test, integration, codegen, run tasks
- stalwart-dev/: IMAP+SMTP dev server reused from sharedinbox
- test/integration/imap_sync_test.dart: login, list mailboxes,
send via SMTP and receive via IMAP
- pubspec.yaml: add flutter_secure_storage
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
