- Add dev-only "Use 10-minute rollover (testing)" toggle in Reminder
Notifications (Account view). Visible only when not on prod API server
(isNotProdServer). Toggle persists and reschedules reminder with
rolloverIntervalMinutes when changed.
- Extend daily notification flow to pass optional rolloverIntervalMinutes
to the plugin: NotificationService/NativeNotificationService options,
PushNotificationPermission dialog options, first-time and edit flows.
- Add settings: reminderFastRolloverForTesting (Settings, AccountSettings,
PlatformServiceMixin boolean mapping, migration 007).
- Centralize isNotProdServer(apiServer) in constants/app.ts; use in
AccountViewView (toggle visibility), ImportAccountView, and TestView.
- Add docs/plugin-spec-rollover-interval-minutes.md for the plugin repo
(configurable rollover interval and persistence after reboot).
Note: Daily notification plugin dependency is currently pointed at the
"rollover-interval" branch for testing this feature.
Resolves long-standing issue where the second scheduled time (after editing
the reminder) did not fire on Android.
- PushNotificationPermission: add open(..., options?: { skipSchedule }).
When skipSchedule is true (edit flow), dialog only invokes callback with
time/message; parent is sole scheduler so the plugin is not called twice.
- AccountViewView: pass { skipSchedule: true } when opening the dialog for
edit; keep cancel (iOS only) + single scheduleDailyNotification in callback.
- NativeNotificationService: serialize scheduleDailyNotification so only one
schedule runs at a time (scheduleLock + doScheduleDailyNotification).
- AccountViewView: guard edit-reminder callback with editReminderScheduleInProgress
so one schedule per user action.
- Gate pre-cancel on Android in edit flow (CONSUMING_APP brief): skip
cancelDailyNotification before schedule on Android; plugin cancels internally.
- Use single stable reminder id and always pass id on both platforms (plugin 1.1.2+).
- Add doc/plugin-android-edit-reschedule-alarm-not-firing.md for plugin repo
(cancel-before-reschedule may cancel the PendingIntent used for setAlarmClock).
- NativeNotificationService: use single stable reminder id on iOS and Android,
always pass id in scheduleOptions (plugin v1.1.2+ optional cleanup).
- Add doc/plugin-fix-scheduleExactNotification-calls.md for plugin repo
(fix Java call sites for scheduleExactNotification 8th parameter).
- package-lock.json: update lockfile.
Android second-schedule issue still present; to be revisited.
- NativeNotificationService: platform-specific schedule/cancel
- iOS: pass id "daily_timesafari_reminder", call cancelDailyReminder before schedule
- Android: no id (plugin uses "daily_notification"), skip pre-cancel to match test app
- Verification: return true when schedule succeeds but reminder not found (avoids error dialog)
- doc: android-daily-notification-second-schedule-issue.md
- Symptom, timing (re-schedule-too-soon), test app vs TimeSafari
- Plugin-side section: entry point, files, likely cause, suggested fixes, repro steps
- For use in plugin repo (e.g. Cursor) to fix second-schedule
Re-scheduled notifications on Android still fail to fire; fix expected in plugin (see doc).
Add native Android components for daily notification plugin integration:
- TimeSafariApplication: Custom Application class to register native fetcher
- TimeSafariNativeFetcher: Implements NativeNotificationContentFetcher interface
- network_security_config.xml: Allow cleartext for local development
Configuration updates:
- AndroidManifest.xml: Link custom Application class, add required permissions
- build.gradle: Add Java 17 compile options and required dependencies
- capacitor.config.ts: Add DailyNotification plugin configuration
- NativeNotificationService.ts: Use "daily_" prefixed ID for schedule rollover
Note: Subsequent notification scheduling after first fire still has issues
that require further investigation.
Co-authored-by: Cursor <cursoragent@cursor.com>
The DailyNotificationReceiver was not being triggered when scheduled alarms
fired, preventing notifications from appearing at the scheduled time.
Changed android:exported from "false" to "true" to allow AlarmManager
broadcasts to reach the receiver, especially when the app is closed or
the device is in doze mode.
This is a work-in-progress change to diagnose why notifications aren't
firing. The receiver should log "DN|RECEIVE_START" when triggered, but
we were not seeing these logs even when alarms were scheduled.
Next steps:
- Test if receiver is now triggered when alarm fires
- Verify notifications appear at scheduled time
- Consider adding permission check if keeping exported=true for security
- Fix permission request to use correct iOS method (requestNotificationPermissions)
- Add robust handling for varying permission result formats
- Fix cancelDailyReminder to pass object parameter matching Swift plugin expectation
- Add notification cancellation before rescheduling to prevent duplicates
- Add verification after scheduling to ensure notification was actually scheduled
- Fix getStatus to handle both array and object response formats
- Enable notifications section in AccountView for native platforms (iOS/Android)
- Add edit button to allow users to modify existing notification time and message
- Add editReminderNotification method with form pre-population
- Add parseTimeTo24Hour helper for time format conversion
Fixes issues where:
- Notifications were stored but not actually scheduled with UNUserNotificationCenter
- cancelDailyReminder failed due to parameter type mismatch
- Notification time updates didn't properly cancel old notifications
- Users couldn't easily edit existing notification settings
The notification section is now visible on native platforms and includes an edit
button that opens the notification dialog with current values pre-populated.
Handle "Connection already exists" error when initializing SQLite database
on Capacitor platforms. The native connection can persist across app
restarts while the JavaScript connection Map is empty, causing a mismatch.
When createConnection fails with "already exists":
- Check if connection exists in JavaScript Map and retrieve it if present
- If not in Map, close the native connection and recreate to sync both sides
- Handle "already open" errors gracefully when opening existing connections
This fixes the issue where clicking "Backup Identifier Seed" would redirect
to StartView instead of SeedBackupView due to database initialization
failures in the router navigation guard.
Fixes navigation issue on both iOS and Android platforms.
- Resolves merge conflicts from master branch integration
- Includes latest features and bug fixes from master
- Maintains clean-db-disconnects branch functionality
Files affected: Multiple components, views, and utilities
Timestamp: Wed Oct 22 07:26:21 AM UTC 2025
- Remove accountDid exclusion from settings object construction in CapacitorPlatformService
- Remove accountDid exclusion from settings object construction in WebPlatformService
- Ensure accountDid is included in retrieved settings for proper DID-specific configuration handling
This change ensures that the accountDid field is properly included when retrieving
settings for the active account, allowing for proper DID-specific configuration
management across both Capacitor (mobile) and Web platforms.
Files modified:
- src/services/platforms/CapacitorPlatformService.ts
- src/services/platforms/WebPlatformService.ts
Timestamp: Wed Oct 8 03:05:45 PM UTC 2025
- Update CapacitorPlatformService.updateDefaultSettings() to use active_identity table instead of hard-coded id=1
- Update CapacitorPlatformService.retrieveSettingsForActiveAccount() to query by accountDid from active_identity
- Add getActiveIdentity() method to CapacitorPlatformService for consistency with WebPlatformService
- Update WebPlatformService.retrieveSettingsForActiveAccount() to match CapacitorPlatformService pattern
- Both services now consistently use active_identity table instead of legacy MASTER_SETTINGS_KEY approach
- Maintains backward compatibility with databaseUtil.ts for PWA migration support
Technical details:
- CapacitorPlatformService: Fixed hard-coded WHERE id = 1 → WHERE accountDid = ?
- WebPlatformService: Fixed retrieval pattern to match new architecture
- Platform services now aligned with migration 004 active_identity table schema
- databaseUtil.ts remains unchanged for PWA-to-SQLite migration bridge
- Update CapacitorPlatformService.updateDefaultSettings() to use active_identity table instead of hard-coded id=1
- Update CapacitorPlatformService.retrieveSettingsForActiveAccount() to query by accountDid from active_identity
- Add getActiveIdentity() method to CapacitorPlatformService for consistency with WebPlatformService
- Update WebPlatformService.retrieveSettingsForActiveAccount() to match CapacitorPlatformService pattern
- Both services now consistently use active_identity table instead of legacy MASTER_SETTINGS_KEY approach
- Maintains backward compatibility with databaseUtil.ts for PWA migration support
Technical details:
- CapacitorPlatformService: Fixed hard-coded WHERE id = 1 → WHERE accountDid = ?
- WebPlatformService: Fixed retrieval pattern to match new architecture
- Platform services now aligned with migration 004 active_identity table schema
- databaseUtil.ts remains unchanged for PWA-to-SQLite migration bridge
- Remove isDevelopment environment checks and migrationLog variable
- Replace conditional logging with consistent logger.debug() calls
- Remove development-only validation restrictions
- Maintain all error handling and warning messages
- Let existing logger handle development mode behavior automatically
This simplifies the migration service logging while preserving all
functionality. The existing logger already handles development vs
production mode appropriately.
- Remove explicit transaction wrapping in migration service that caused
"cannot start a transaction within a transaction" errors
- Fix executeSet method call format to include both statement and values
properties as required by Capacitor SQLite plugin
- Update CapacitorPlatformService to properly handle multi-statement SQL
using executeSet for migration SQL blocks
- Ensure migration 004 (active_identity_management) executes atomically
without nested transaction conflicts
- Remove unnecessary try/catch wrapper
Fixes iOS simulator migration failures where:
- Migration 004 would fail with transaction errors
- executeSet would fail with "Must provide a set as Array of {statement,values}"
- Database initialization would fail after migration errors
Tested on iOS simulator with successful migration completion and
active_identity table creation with proper data migration.
- Single SQL source: Define MIG_004_SQL constant to eliminate duplicate SQL definitions
- Atomic execution: Add BEGIN IMMEDIATE/COMMIT/ROLLBACK around migration execution
- Name-only check: Skip migrations already recorded in migrations table
- Guarded operations: Replace table-wide cleanups with conditional UPDATE/DELETE
Changes:
- migration.ts: Extract migration 004 SQL into MIG_004_SQL constant
- migration.ts: Use guarded DELETE/UPDATE to prevent accidental data loss
- migrationService.ts: Wrap migration execution in explicit transactions
- migrationService.ts: Reorder checks to prioritize name-only skipping
Benefits:
- Prevents partial migration failures from corrupting database state
- Eliminates SQL duplication and maintenance overhead
- Maintains existing APIs and logging behavior
- Reduces risk of data loss during migration execution
Test results: All migration tests passing, ID generation working correctly
- Phase 1: Simplify Migration Definition ✅
* Remove duplicate SQL definitions from migration 004
* Eliminate recovery logic that could cause duplicate execution
* Establish single source of truth for migration SQL
- Phase 2: Fix Database Result Handling ✅
* Remove DatabaseResult type assumptions from migration code
* Implement database-agnostic result extraction with extractSingleValue()
* Normalize results from AbsurdSqlDatabaseService and CapacitorPlatformService
- Phase 3: Ensure Atomic Execution ✅
* Remove individual statement execution logic
* Execute migrations as single atomic SQL blocks only
* Add explicit rollback instructions and failure cause logging
* Ensure migration tracking is accurate
- Phase 4: Remove Excessive Debugging ✅
* Move detailed logging to development-only mode
* Preserve essential error logging for production
* Optimize startup performance by reducing logging overhead
* Maintain full debugging capability in development
Migration system now follows single-source, atomic execution principle
with improved performance and comprehensive error handling.
Timestamp: 2025-09-17 05:08:05 UTC
- Fix multi-statement SQL execution issue in Capacitor SQLite
- Add individual statement execution for migration 004_active_identity_management
- Implement automatic recovery for missing active_identity table
- Enhance migration system with better error handling and logging
Problem:
Migration 004 was marked as applied but active_identity table wasn't created
due to multi-statement SQL execution failing silently in Capacitor SQLite.
Solution:
- Extended Migration interface with optional statements array
- Modified migration execution to handle individual statements
- Added bootstrapping hook recovery for missing tables
- Enhanced logging for better debugging
Files changed:
- src/services/migrationService.ts: Enhanced migration execution logic
- src/db-sql/migration.ts: Added recovery mechanism and individual statements
This fix ensures the app automatically recovers from the current broken state
and prevents similar issues in future migrations.
- Consolidate migrations 004, 005, and 006 into single 004_active_identity_management
- Remove redundant migrations 005 (constraint_fix) and 006 (settings_cleanup)
- Implement security-first approach with ON DELETE RESTRICT constraint from start
- Include comprehensive data migration from settings.activeDid to active_identity.activeDid
- Add proper cleanup of orphaned settings records and legacy activeDid values
- Update migrationService.ts validation logic to reflect consolidated structure
- Fix migration name references and enhance validation for hasBackedUpSeed column
- Reduce migration complexity from 3 separate operations to 1 atomic operation
- Maintain data integrity with foreign key constraints and performance indexes
Migration successfully tested on web platform with no data loss or corruption.
Active DID properly migrated: did:ethr:0xCA26A3959D32D2eB5459cE08203DbC4e62e79F5D
Files changed:
- src/db-sql/migration.ts: Consolidated 3 migrations into 1 (-46 lines)
- src/services/migrationService.ts: Updated validation logic (+13 lines)
- Fix syntax error in logger.ts: change 'typeof import' to 'typeof import.meta'
to resolve ESBuild compilation error preventing web build
- Align CapacitorPlatformService.insertNewDidIntoSettings with WebPlatformService:
* Add dynamic constants import to avoid circular dependencies
* Use INSERT OR REPLACE for data integrity
* Set proper default values (finishedOnboarding=false, API servers)
* Remove TODO comment as implementation is now parallel
- Fix Playwright test timing issues in 60-new-activity.spec.ts:
* Replace generic alert selectors with specific alert type targeting
* Change Info alerts from 'Success' to 'Info' filter for proper targeting
* Fix "strict mode violation" errors caused by multiple simultaneous alerts
* Improve test reliability by using established alert handling patterns
- Update migrationService.ts and vite.config.common.mts with related improvements
Test Results: Improved from 2 failed tests to 42/44 passing (95.5% success rate)
Build Status: Web build now compiles successfully without syntax errors
- Update migration 003 to match master deployment (hasBackedUpSeed)
- Rename migration 004 for active_identity table creation
- Update migration service validation for new structure
- Fix TypeScript compatibility issue in migration.ts
- Streamline active identity upgrade plan documentation
- Ensure all migrations are additional per team guidance
Migration structure now follows "additional migrations only" principle:
- 003: hasBackedUpSeed (assumes master deployment)
- 004: active_identity table with data migration
- iOS/Android compatibility confirmed with SQLCipher 4.9.0
Files: migration.ts, migrationService.ts, active-identity-upgrade-plan.md
- Fix $getActiveIdentity() logic flow preventing false "empty table" warnings
- Implement auto-selection of first account when activeDid is null after migration
- Consolidate 003 and 003b migrations into single 003 migration
- Re-introduce foreign key constraint for activeDid referential integrity
- Add comprehensive debug logging for migration troubleshooting
- Remove 003b validation logic and update migration name mapping
Fixes migration from master to active_did_redux branch and ensures system
always has valid activeDid for proper functionality.
- Fix line breaks and indentation for long SQL queries
- Improve readability of error message formatting
- Remove trailing whitespace and standardize spacing
- Apply consistent formatting to active_identity table validation logic
The complex table rewrite approach in migration 003_active_did_separation was
failing on iOS SQLite, causing "no such table: active_identity" errors. The
migration was being marked as applied despite validation failures.
Changes:
- Simplify migration SQL to only create active_identity table and migrate data
- Remove complex table rewrite that was failing on iOS SQLite versions
- Remove foreign key constraint that could cause compatibility issues
- Update validation logic to focus on active_identity table existence only
- Remove validation check for activeDid column removal from settings table
This approach is more reliable across different SQLite versions and platforms
while maintaining the core functionality of separating activeDid into its own
table for better database architecture.
Fixes iOS build database errors and ensures migration completes successfully.
- Remove problematic $ensureActiveIdentityPopulated() that auto-selected identities
- Add user-friendly $needsActiveIdentitySelection() and $getAvailableAccountDids() methods
- Fix missing updateActiveDid implementation in CapacitorPlatformService
- Resolve race condition in HomeView initialization causing feed loading failures
- Improve TypeScript error handling in ContactsView invite processing
Addresses team concerns about data consistency and user control for identity selection.
- Update PlatformServiceMixin interface to include $getActiveIdentity
- Improve apiServer default handling across all platforms
- Add better error handling for platform service methods
- Ensure consistent behavior across web and electron platforms
- Change server switching logs from info to debug level
- Implement structured error logging for profile CRUD operations
- Handle HTTP status codes 400, 401, 403, 404, 409 gracefully
- Suppress full error stack traces for expected API responses
- Maintain user notifications while improving console readability
- Add timestamp and context to all profile-related error logs
Improves developer experience by reducing console noise while preserving
debugging information and user-facing error handling.
Removes over-engineered ProfileService and ServiceInitializationManager
classes that were only used in one place. Inlines all profile logic
directly into AccountViewView.vue to reduce complexity and improve
maintainability.
- Deletes ProfileService.ts (325 lines)
- Deletes ServiceInitializationManager.ts (207 lines)
- Inlines ProfileData interface and methods into AccountViewView
- Maintains all existing functionality while reducing code footprint
perf(logging): convert excessive info logs to debug level
Reduces console noise by converting high-frequency, low-value logging
from info to debug level across navigation, API calls, and component
lifecycle operations. Improves performance and reduces log verbosity
for normal application flow.
- Router navigation guards: info → debug
- Plan loading operations: info → debug
- User registration checks: info → debug
- Image server rate limits: info → debug
- Component lifecycle events: info → debug
- Settings loading operations: info → debug
Maintains warn/error levels for actual issues while reducing noise
from expected application behavior.
- Revert ProfileService from broken /api/partner/userProfile endpoint to working /api/partner/userProfileForIssuer/${did}
- Fix location data display by restoring single profile object response parsing
- Remove complex array handling logic that was unnecessary for current user profiles
- Restore original working functionality that was broken by recent refactoring
Problem: Recent ProfileService creation changed endpoint from working userProfileForIssuer/${did}
to broken userProfile (list endpoint), causing location data to not display properly.
Solution: Revert to original working endpoint and response parsing logic that returns
single profile objects with location data instead of arrays of all profiles.
Files changed:
- src/services/ProfileService.ts: Restore working endpoint and simplify response parsing
Testing: Profile loading now works correctly for both existing and new profiles,
location data is properly extracted and displayed, maps render correctly.
