Integration testing for role switch end-to-end flow
epic-role-based-access-control-ui-and-navigation-task-013 — Write integration tests covering the full role switch flow: authenticate as multi-role user → verify home screen renders correct variant → open Role Switch Widget → select alternate role → verify home screen transitions → verify bottom navigation tabs update → verify route guard enforces new role permissions → switch back and confirm state restoration.
Acceptance Criteria
Technical Requirements
Execution Context
Tier 7 - 84 tasks
Can start after Tier 6 completes
Implementation Notes
The key complexity here is waiting for asynchronous BLoC state transitions before asserting UI state. Use await tester.pumpAndSettle() after each action, but set a generous timeout (3–5s) to avoid false failures from animation completion. For route guard enforcement, navigate programmatically using the GoRouter test helper or by finding and tapping a link widget that leads to a role-restricted route — do not call router.go() directly in the test as this bypasses the guard. Seed the test user fixture at the start of the test run using a beforeAll-equivalent setup.
Use Riverpod's ProviderContainer overrides to inject a mock AuthRepository so the test does not depend on a live Supabase instance. Pay special attention to StatefulShellRoute tab state preservation: after a role switch, tabs that existed in both roles should not lose their navigation stack.
Testing Requirements
Use Flutter integration_test package (not flutter_test widget tests) to drive a real app instance. Create a test fixture factory that seeds a multi-role user with at least two roles (e.g., peer_mentor + coordinator). Structure tests as a sequential scenario using testWidgets with WidgetTester pump-and-settle after each BLoC state emission. Mock Supabase calls using a local Supabase instance or an HTTP mock layer (e.g., http_mock_adapter) to avoid flakiness.
Include at least: (1) happy-path role switch, (2) switch-back state restoration, (3) route guard enforcement after switch, (4) rapid double-tap on Role Switch Widget (should not cause duplicate transitions). Run on both platforms in CI.
Combining GoRouter's declarative redirect logic in the route guard with StatefulShellRoute's stateful branch management is known to produce subtle bugs where the shell rebuilds unnecessarily on role switches, losing tab state or causing double-navigation events.
Mitigation & Contingency
Mitigation: Implement the route guard as a GoRouter redirect callback that only evaluates role from an already-resolved Riverpod provider (not async). Use a dedicated ShellRoute navigator key per tab branch to anchor state independently of role-driven rebuilds. Write integration tests for the full navigation graph.
Contingency: If StatefulShellRoute state loss is confirmed during QA, fall back to a manual tab state preservation approach using a TabStateManager service that caches the last route per tab and restores it after role switches, decoupling tab state from the shell lifecycle.
The role-based home screen must render three significantly different layouts (coordinator dashboard, peer mentor activity summary, org admin overview). If these variants are implemented as a single widget with conditionals, the file will become unmaintainable and difficult to test in isolation, especially as each variant grows with downstream feature additions.
Mitigation & Contingency
Mitigation: Design the role-based home screen as a router/dispatcher widget that delegates to three separate variant widgets (CoordinatorHomeView, PeerMentorHomeView, OrgAdminHomeView). Each variant is independently testable and can be developed by separate team members in parallel.
Contingency: If variant coupling has already occurred before this risk is addressed, refactor to the dispatcher pattern in a dedicated cleanup task before feature handoff. The dispatcher pattern is a straightforward extraction that carries low refactoring risk.
The no-access screen must link global admin users to the correct admin portal URL, which may differ per organization (NHF, HLF, Blindeforbundet each have their own admin portals). Hardcoding a single URL will result in wrong or broken links for some global admin users.
Mitigation & Contingency
Mitigation: Source the admin portal URL from the organization's configuration record in Supabase rather than hardcoding it. The no-access screen reads the active org context and resolves the portal URL dynamically. Provide a safe fallback to a generic Norse Digital Products support page if the URL is not configured.
Contingency: If dynamic URL resolution is not ready when the no-access screen ships, display a static instruction to contact the organization's administrator along with a support email address as an interim measure, and track the URL configuration task as a follow-up.