Changelog for discord-haskell-voice

Unreleased changes

3.0.0 — 2025 March

• Breaking API changes

  • Replaced previously separate playing functions with a single play function
  • New AudioResource type represents a resource to be played, made with e.g. createYoutubeResource
  • New audio transformation abilities via FFmpeg flags and Haskell conduits
  • Use FFmpeg's direct Ogg/Opus output to skip the FFmpeg->PCM->Opus translation if there are no Haskell conduits that operate on PCM
  • Can now specify a codec in play to skip on using FFmpeg at all if input is already PCM or Opus
  • Can now use ffprobe to automatically detect if the input resource is already PCM or Opus and intelligently skip FFmpeg
  • New dependency on typed-process for safer external process
  • New dependency on opus from Hackage instead of direct git source
  • Remove ExceptT from the Voice monad stack
  • Remove SubprocessException type from the library as it became unused
  • Remove InvalidPayloadOrder from the VoiceError ADT since it became unused

• Bug Fixes

  • Fix OpusBufferTooSmall when receiving Opus data due to buffer being half the size it should be
  • Fix the library crashing when joining a call with another user already in it, which triggers Opcodes 11/18/20 before Opcode 4
  • Fix leave causing MVar thread deadlock due to the BoundedChan consumer thread being killed
  • Fix possibility of audio artifacts when sending multiple audio resources back-to-back due to incorrect silence frames #45
  • Fix posibility of the last few bytes of Opus frames being cut off during transmission due to a buffer too small #45

• Miscellaneous

  • Support GHC 8.10.7, 9.0.2, 9.2.4, and 9.6.6
  • Improve BasicMusicBot example to be less lisp-y in terms of brackets, and fix all warnings
  • Improve error messages when parsing Ogg containers fail
  • Use DerivingStrategies in the library code to make explicit where deriving typeclasses are from
  • Add a compile flag (-use-crypton) to use a crypton-based encryption backend, which removes the necessity for libsodium
  • Remove containers dependency
  • Relax package bounds:
    • aeson from ==1.5.6.0 to <2.3
    • bytestring from <0.11 to <0.13
    • conduit from <=1.3.4.2 to <1.4.0.0
    • mtl from ==2.2.2 to <2.4
    • saltine from <0.2 to <0.4
    • stm from <2.5.1 to <2.6
    • text from <2 to <3
    • time from <=1.13 to <1.15
    • websockets from <0.12.8 to <0.14
    • wuss from <=1.2 to <2.1
    • discord-haskell from <= 1.14.0 to <= 1.17.1
  • Add build CI for Cabal & Stack for all supported GHC versions except 9.6.6
  • Renamed master branch to main
  • Updated copyright to current year and include contributors where applicable
  • Added link to GitHub Sponsors
  • Updated Haddock for all functions and added examples and usage to many functions
  • Added upper bound of <0.5 to microlens and microlens-th

2.3.1 — 2022 July

• Update discord-haskell dependency bounds to >= 1.12.0 && <= 1.14.0.
• Use UnliftIO.MVar functions internally for MVar operations in DiscordHandler
• IOExceptions thrown by e.g. createProcess during runVoice are no longer caught and subdued - they are propagated to the user.

2.3.0 — 2022 May

• Export playYouTubeWith and playYouTubeWith' from Discord.Voice.
• Update discord-haskell dependency bounds to >= 1.12.0 && <= 1.13.0.
• Migrate from lens to microlens, following the opus package doing the same.

2.2.2 — 2022 February

• Update discord-haskell dependency to 1.12.0
• Bump copyright to 2022
• Fix incomplete pattern match crash in the example bot when using bot --bash-completion-script

2.2.1 — 2022 January

• Patch README having incorrect non-published details after Hackage publication.

2.2.0 — 2022 January

• Change the definition of Voice from a type alias exposing dangerous internal handles, to a newtype wrapper. This also changes the definition of liftDiscord to maintain identical behaviour.
• Update discord-haskell dependency to 1.11.0

2.1.0 — 2022 January

• Removed updateSpeakingStatus from the publicly exported function list for Discord.Voice.

2.0.0 — 2021 December

• Rewrite the entire library (see #1).
• Introduce the Voice monad, and all functions in it: join, play, and all other variants of play.
• Add lens as a dependency for internal library use.
• Add conduit as the main method of piping and transforming audio on the fly.
• Remove all previous functions: joinVoice, leaveVoice, playPCM, etc.
• Add package documentation to public modules, and make sure the abstraction layer is solid (don't export useless internals).
• Rename the JoinSpecificVC example to BasicMusicBot and add a bot volume command to change the volume.

0.0.1 — 2021 June

• Initial release.
• Implement joinVoice, leaveVoice, etc and use DiscordVoiceHandle to maintain a reference to the voice handle.
• Add JoinAllVC and JoinSpecificVC as example usages of the library.