- Introduce PeerWatermarkStore to track max HLC received per peer for
shared resource incremental sync - Initialize the
peer_received_watermarks table at startup - Persist received watermarks
during backfill and after backfill completion - Query the max watermark
across peers via PeerWatermarkStore - Expose the new module and export
PeerWatermarkStore from sync
Process the pre-sync current_state data in dependency order before
applying HLC-ordered log entries. This ensures base records exist for FK
constraints, and then resolves dependent changes after inserting
snapshot records. Extend FK error handling to recognize foreign key
failures and extract missing UUIDs for buffering and retry.
Skip self-watermarks when peer is the local device; log a warning. Apply
current_state in dependency order by computing registry order. Fall back
to unordered if the computation fails. On automatic backfill failure,
reset device state to Uninitialized. This triggers a retry and clears
backfill_attempted.
- Introduce deterministic UUID generation for library defaults (spaces,
groups, and items) - Add post-backfill rebuild support in the sync
registry (with_rebuild) - Export BatchFkMapResult type via the sync API
- Track per-peer RTT latency via SyncMetricsCollector integration - Use
deterministic IDs when creating default space and related items in
LibraryManager
- Revert to the November 16 baseline for realtime sync - Use a single
state watermark and drop per-resource watermarks, counts, and hashes
from watermark messages - Revert DataAvailableNotification handling and
the updated FK filtering approach in favor of NULL-based handling - Add
registry helper get_model_type_by_table to map tables to models for FK
resolution - Update mocks and tests to reflect the baseline (linkage
stats in summary) - Fix test infrastructure: remove block_in_place usage
and minor imports - Remove obsolete test fixture file
Introduce inventory-based registration for Syncable models. Add
SyncableInventoryEntry and macros to register models to register models
at link time. Refactor registry and FK mapper to use the registry for FK
resolution and lookups. Add post-backfill hooks and per-model FK
callbacks. Register all entities via the new macros.
- Remove relay-URL based pairing flow and rely on pkarr/DNS for remote
discovery - Extend pairing code with optional node_id and adopt version
2 QR format - Generate pairing codes using node_id and drop relay_url in
non-QR flows - Enable combined mDNS, pkarr, and DNS discovery in the
endpoint - Cache relay_url in persisted device info for reconnection
optimization - Update protocol handlers to propagate node_id and
pkarr-based flow - Adjust tests and logging to reflect the new
cross-network pairing
- Add apply_shared_change handlers for Space, SpaceGroup, and SpaceItem
to upsert by UUID and delete by UUID
- Update registry to use shared_with_query and call apply_shared_change;
remove deletion hooks
- Enhance SyncMonitorPopover to display a colored state badge in a
header bar
- Wire library manager into VolumeManager to enable merging tracked
volumes from the database into runtime state - Validate visibility
before tracking: reject non-user-visible volumes - Merge DB metadata for
tracked volumes (library_id, display_name) into detected volumes and
emit ResourceChanged for user-visible ones - Emit ResourceChanged on
adds and updates; emit ResourceDeleted on removals for user-visible
volumes to support UI reactivity - Generate stable UUIDs for new volumes
from fingerprints (APFS and macOS)
1. Watermark Messages: - Nov 16: Simple - single my_state_watermark,
single state_watermark in response - Main: Complex - per-resource
my_resource_watermarks HashMap, resource counts, content hashes 2.
DataAvailableNotification: - Nov 16: Doesn't exist - Main: New message
type that triggers watermark exchanges 3. FK Mapper: - Nov 16: Sets
missing FKs to NULL, applies all records - Main: Marks records with
missing FKs as failed, filters them out 4. Watermark Storage (internal,
not in messages): - Nov 16: Single last_watermark per resource - Main:
Dual watermarks - cursor_watermark and
- Add show_paired flag to device listing (CLI arg and input) - Extend
LibraryDeviceInfo with is_paired and is_connected; include Paired
devices in library listing when requested - Add Devices group to UI
(DevicesGroup) and hook into SpaceGroup - Extend device queries/types to
support show_paired and paired devices - Refactor Dockerfile to
multi-stage Bun + Rust builds; reuse web assets - Remove obsolete
core/ops/entries/mod.rs
- Introduce an Axum-based HTTP server with an embedded daemon and a
JSON-RPC proxy to the daemon via a Unix socket - Bundle web UI assets
into the server with an assets feature and a build.rs that builds the
frontend using pnpm - Add multi-stage Dockerfile, docker-compose.yml,
and a Distroless runtime image - Provide TrueNAS deployment support with
a build script and setup guide - Add a new web UI (apps/web) with a
Vite-based dev/build flow and a web platform shim for the frontend -
Implement server logic (apps/server/src/main.rs): health, auth, /rpc
proxy and data-dir/socket-path wiring - Include server-specific
Cargo.toml and a comprehensive server README - Add architecture and
memory-focused docs to guide usage and design - Minor core tweak:
simplify location/resource event emission in
core/src/location/manager.rs to align with new flow - Tauri app: adjust
menus to add an Edit submenu and remove unused items
- Scope the closure query to the indexing path by resolving its entry_id
from directory_paths and use that for descendants; fall back to the
location root if not found. - Seed all ancestors between the location
root and the indexing path into the in-memory cache during the
processing phase to ensure correct parent links. - Seed ancestor
directories into the create path cache to prevent ghost folders on
subpath indexing. - Add test core/tests/indexing_move_folder_bug_test.rs
reproducing the ghost-folder bug when moving a folder into a managed
location and asserting correct parent linkage.
- Updated file type checks from `file.kind.type` to `file.kind` for consistency across various components in the Explorer views.
- Enhanced the `Thumb` component to conditionally hide the icon based on thumbnail loading status.
- Adjusted the `HeroStats` component for improved readability and structure.
- Added a new `iconScale` prop to the `FileInspector` component's thumbnail for better visual scaling.
- Introduce useJobCount to track active and running jobs via query and
event subscriptions. - Refactor JobManagerPopover to render a separate
content component and rely on the new hook for counts. - Add
JobManagerPopoverContent to render filtered JobList with pause/resume
callbacks and dynamic height. - Update useJobManager to subscribe to job
events via a single filtered subscription, replacing per-event handlers.
- Extend client types to support event_type filtering for subscriptions.
- Tidy up UI logs and minor rendering tweaks in QuickPreview and
ContentRenderer, including folder icon handling. - Adjust storage
overview to use real volume fields and drop dummy data paths
- Updated the subproject commit reference to indicate a dirty state.
- Added blank lines to various files for improved readability and consistency across the codebase.
- Added `include_descendants` option to event filters, allowing recursive path matching for resource events.
- Updated `affects_path` method to support descendant matching, improving event handling accuracy.
- Refactored subscription logic to utilize the new filtering capabilities, ensuring only relevant events are processed.
- Introduced tests for event filtering to validate exact vs. recursive matching, enhancing reliability of event-driven updates.
- Updated related components to leverage the new filtering options, improving overall performance and user experience.
- Added a new "Open Library" menu item to the application, allowing users to select a library folder.
- Implemented the backend logic to handle library opening, including JSON-RPC requests to the daemon.
- Introduced a filesystem watcher in the LibraryManager to monitor library directory changes, enabling automatic opening and closing of libraries based on filesystem events.
- Enhanced error handling for library operations and added relevant logging for better traceability.
- Updated the LibraryError enum to include filesystem watcher errors, improving error reporting.
- Introduced a new constructor for the HLC struct to allow explicit initialization with timestamp, counter, and device ID.
- Updated indexing tests to improve error handling for entry IDs and added a comprehensive test for reindexing after offline changes, ensuring all original and new files are detected correctly.
- Introduced a new ProxyPolicy struct to manage video scrubbing settings, including options for enabling and regenerating proxy files.
- Updated JobPolicies to include the new proxy policy with default values.
- Enhanced FileInspector to filter batch events by resource ID for improved event handling.
- Modified useNormalizedCache to skip path validation checks when resourceId is provided, optimizing resource handling.
- Updated the LocationUpdateAction to clone the location before building the ActiveModel, ensuring data integrity during updates.
- Implemented a new event emission for ResourceChanged, providing real-time updates to the UI upon location changes.
- Enhanced error handling for missing entry IDs and directory paths, improving robustness in the update process.
- Refactored the LocationInspector component to include new quick actions for reindexing locations, enhancing user interaction.
- Removed redundant isGlobalList flags from various components to streamline the codebase.
- Added critical safety checks to validate that indexing paths are within specified location boundaries, preventing potential data loss from incorrect event routing.
- Improved event routing logic in the LocationWatcher to match events to the longest valid location path, ensuring accurate event handling in scenarios with multiple locations.
- Enhanced logging for successful path validation and event routing, providing better insights into the processing flow.
- Introduced unit tests to verify path matching functionality across multiple locations, ensuring robustness in event handling.
- Implemented a migration step in the LibraryManager to ensure the database schema is up to date before proceeding with operations.
- This change enhances the reliability of the application by preventing potential schema-related issues during runtime.
- Updated the subproject commit reference to indicate a dirty state.
- Added blank lines to various files for improved readability and consistency across the codebase.
