Changelog

This page keeps a detailed human friendly rendering of what’s new and changed in specific versions.

v1.3.0 - 2026-07-20

Added

• Added various helper methods to Queue:

  • Queue.sort() to sort the queue by a given key function. Similar to Python’s built-in list.sort() method.

    • Example: queue.sort(key=lambda track: track.title) to sort by track title.

  • Queue.reverse() to reverse the order of the queue.

  • Queue.dedupe() to remove duplicate tracks from the queue.

  • Queue.__reversed__() to allow the built-in reversed() function on the queue.

• Added History.enabled property for checking whether history recording is active.

• Added Player.skip_to() to skip directly to a track at a specific queue index.

• Added Queue.move() to move a track from one index to another, shifting other tracks accordingly.

• Added Queue.remove_at() to remove and return a track by index without side effects on the current track.

• Added a key parameter to Queue.remove() and Queue.remove_wait() for removing tracks by an arbitrary derived value instead of equality, without needing to hold the original Playable instance.

  • Example: queue.remove([track.identifier], key=lambda t: t.identifier) to remove a track by its identifier.

• Added ShuffleMode and Queue.shuffle_mode for persistent shuffle support.

  • Independent of QueueMode.

  • While ShuffleMode.PERSISTENT is active, Queue.get() pops a random track instead of the next one in order, without reordering the underlying queue. Switching back to ShuffleMode.DEFAULT resumes playback in the original order.

  • Queue.shuffle() is unaffected and still performs its usual one-time in-place reorder.

• Added a key parameter to Queue.get() and Queue.get_wait() to avoid picking a track that collides with the current track on an arbitrary derived value, e.g. to avoid repeats.

  • Example: queue.get(key=lambda track: track.author) to avoid picking a track by the currently playing artist.

  • If every remaining candidate collides, the collision is allowed rather than raising.

• Added a key parameter to Player.skip(), forwarded straight through to Queue.get(), so the same repeat-avoidance behaves consistently whether tracks are pulled from the queue directly or via the player.

  • Example: await player.skip(key=lambda track: track.author).

Fixed

• Fixed Queue.get_wait() not registering the waiter before attempting to retrieve a track, causing a race condition where a concurrent put() could be missed and the waiter would hang indefinitely.

• Fixed Queue.history returning a History instance when history is disabled; now correctly returns None.

• Fixed Player.play() not pushing the previously playing track to history before starting a new track, matching the documented behavior.

• Fixed Player.stop() with clear_queue=True not clearing AutoPlay-discovered tracks from the queue.

• Fixed Player.stop() silently resetting Queue.mode to QueueMode.NORMAL when clear_queue=True; this is now documented behavior.

• Fixed Queue.copy() creating a new History that shared a mutable reference to the original settings object, causing mutations to enabled or max_items to leak between copies.

• Fixed Queue.remove() with a key parameter ignoring duplicate lookup values and crashing on unhashable types; both the remove_all=True and remove_all=False paths now consistently respect user-provided values.

• Added missing exports for AutoPlaySeedMissing, FrameworkClientMismatch, and FrameworkImportError to the top-level __all__.

• Fixed Player.on_track_end() dispatching TrackEndEvent with the wrong original track when skip() was triggered, because _original_track was overwritten by play() before the event was emitted.

• Fixed Player.previous() pushing the previous track back into history after play() was called, creating a duplicate history entry.

• Fixed Filters constructor not calling super().__init__(), leaving _client and _data uninitialised on direct construction.

Miscellaneous

• Consolidated the separate pr-checks and pr-template workflows into a single pr-governance workflow.

• Scoped all CI job permissions to read-only and disabled credential persistence on checkout for improved security.

• Added a 15-minute timeout to every CI job.

• Enabled pip caching across CI workflows.

• Added concurrency control to the lint workflow to cancel redundant runs.

• Added twine check and build provenance attestation to the release workflow.

• Hardened the template-check validation and excluded draft PRs from template enforcement, triggering checks only on ready_for_review.

• Added sticky h2/h3 headers, a hover-highlights toggle persisted in localStorage, and alphabetical “On this page” TOC sorting to the docs.

• Enabled nitpicky Sphinx mode and sphinx-autodoc-typehints; replaced raw :class: references with proper cross-references throughout.

• Exposed missing exception classes (SonoLinkException, NodeError, NetworkError) and Version/GitInfo models in the API docs.

v1.2.1 - 2026-07-04

Added

• Added the missing NetworkError export to the public API.

Fixed

• Fixed a typo in __repr_attrs__ of WebSocketClosedEvent.

• Fixed an incorrect datetime timestamp in time caused by differing time units.

• Fixed an issue where the node status would remain stuck on NodeStatus.CONNECTING after an attempt_connect failure.

• Fixed Filters not populating data on the instance.

• Fixed a race condition in the node reconnection flow that caused an intermittent “Timed out waiting for node READY payload” error after prolonged uptime.

v1.2.0 - 2026-06-23

Added

• Added Node.is_connecting property for checking whether a node is currently establishing a connection.

• Added the missing auto_reconnect parameter to Client.create_node() for controlling whether the client should automatically attempt to reconnect to the node on disconnection. This defaults to True.

• Added Node.reconnect() for manually triggering a reconnection to the node.

• Added original and original to access the original track passed to Player.play().

• Serializable extras are now forwarded to Lavalink as user_data and accessible via extras on reconstructed tracks. Non-serializable values are warned and accessible via .original.extras.

• Added support for region-aware node selection:

  • Added NodeRegion enum representing the available Discord voice channel regions a node can serve.

  • Added regions parameter to Client.create_node() and Node to associate a node with one or more NodeRegion values (or raw strings).

  • Added region parameter to Client.get_best_node(), Client.search_track(), Client.decode_track(), and Client.decode_tracks() to drive region-aware node selection.

  • BasePlayer now automatically passes its voice region to Client.get_best_node() when selecting a node on connect.

Fixed

• Fixed an issue where the client would not attempt to reconnect to a node after a disconnection, even if auto_reconnect was enabled.

• Fixed an issue with using the speed extra, or the curl-cffi package where it would raise a CurlError and the library would fail to handle it properly, causing the node to be left in a broken state.

• Fixed an internal race condition where Node.connect() could be called on an already-connected node, causing a RuntimeError to be silently raised in the background after extended uptime.

v1.1.1 - 2026-05-15

Fixed

• Recursion issue in sonolink.Player custom subclass and instance checks that raised a RecursionError during issubclass or isinstance checks.

• sonolink.Player not properly disconnecting after Discord 4014 and 4022 voice close events, which could leave players stuck in an invalid voice session state.

• sonolink.Player not disconnecting on stale LavaLink session.

• Missing typings for sonolink.gateway.PlayerDisconnectEvent.extra_data on sonolink.gateway.PlayerDisconnectEvent.

v1.1.0 - 2026-05-14

Added

• Added support for Nextcord via NextcordPlayer.

  See the Nextcord examples for usage.

• Added sonolink.get_client for retrieving the active client instance.

• Added Client.get_node() to retrieve a specific node from the client.

• Added three new events:

  • on_sonolink_websocket_closed() (WebSocketClosedEvent)

  • on_sonolink_stats_receive() (StatsEvent)

  • on_sonolink_player_disconnect() (PlayerDisconnectEvent)

• Added Player.update() to quickly update multiple player settings at once. Player.autoplay_settings, Player.history_settings, and Player.queue_mode are now exposed and can be set directly.

• Added TrackSourceType.DEEZER as a supported track source.

• Added Queue.count and History.count properties.

• Added Queue.autoplay_tracks and Queue.put_autoplay() to fix autoplay queue priority via a dedicated stream. autoplay is also now exposed.

• Added AutoPlaySeedMissing, a specific error raised when autoplay has no seed tracks to generate from.

• Added framework-specific import and mismatch exceptions:

  • FrameworkClientMismatch

  • FrameworkImportError

• Added a Getting Started guide.

• Added four new migration guides:

  • Mafic

  • Lavalink.py

  • Lavaplay.py

  • Pomice

  The existing WaveLink guide has also been updated.

• Added a py.typed marker file for improved typing compatibility with type checkers.

• Added two new keyword arguments to Client.create_node(): host and port for specifying the node URI in parts as an alternative to the single uri string; uri is now optional to allow for this.

• Added a Code of Conduct for project community guidelines.

Changed

• WSCloseEvent has been renamed to WebSocketClosedEvent.

• get_best_node() now prefers nodes with known stats over nodes with no reported stats.

• Swapped asyncio.TimeoutError with the built-in TimeoutError throughout.

• Cleaned up documentation for all enums.

Removed

• Removed the fallback to discord.py as the default framework.

• Removed the unused value argument from Player.pause(). Player.resume() is no longer an alias for unpausing — it must be called explicitly to resume playback.

Fixed

• Fixed incorrect generics on the PycordPlayer.

• Fixed force parameter handling in Player.disconnect.

• Fixed the library raising QueueEmpty when skipping tracks for the user internally.

Miscellaneous

• sonolink.gateway.Node has been split into multiple components internally for better separation of concerns and maintainability. This is not a breaking change as the public API remains the same, but it should improve code readability and future extensibility.

• Formatted the project with Ruff for consistent code style and linting.

v1.0.1 - 2026-04-12

Fixed

• Inconsistent environment variable usage across code and documentation

• Fixed performance and memory issue in PlayerFactory caused by repeated metadata scans:

  • Reduced connection time (~11x faster)

  • Reduced memory peak (~12x lower)

  • Eliminated repeated expensive importlib.metadata.packages_distributions() calls

v1.0.0 - 2026-04-11

Initial release. For more information on what this added, consider referring to the API Reference or the examples.