Skip to content

Latest commit

 

History

History
64 lines (53 loc) · 3.71 KB

roadmap.md

File metadata and controls

64 lines (53 loc) · 3.71 KB
Raw

Roadmap

Most roadmap tasks should be available here: https://github.com/beefchimi/earwurm/issues

While we are intentionally keeping the scope of Earwurm to an absolute minimum, there are still some features / tasks that remain to be completed.

Considerations

  1. Should we make context available to the consumer so that they can re-use it for other things?
    • Can cause many complications within Earwurm.
  2. Should we have some “SSR/node detection” to know if we can actually use the Web Audio API?
    • Can the Web Audio API work on a server / in node?
  3. Should we offer a utility to help detect:
    • AudioContext support.
    • webm format support.
    • opus codec support.
  4. Should we implement the “distortion fix” within the unlock method?
    • Howler uses this, but I'm not sure if its relevant for evergreen browsers.
  5. Consider a autoSuspendDelay config option.
    • Accepts a ms value that defaults to 30000.
    • If 0 is passed to autoSuspendDelay, then we consider that a “opt-out” value and disable the auto-suspension.
  6. Should we allow “restarting” a Earwurm instance after .teardown() has been called?
    • My current preference is, once .teardown() is called, you are done with that instance and will not be resurrecting it.
    • We are not however setting #context and managerGain to null, because that would introduce a null type that I do not want to check for within the code.
  7. Investigate if it makes any sense to utilize a OfflineAudioContext.
  8. Consider a delay property for a Sound.
    • The idea is that the delay is persistent for that sound.
    • The delay is included in each subsequent loop, instead of ONLY applying to the first iteration.
    • Maybe a startPad and endPad make sense?
  9. Should we offer a way to "pre-fetch" all audio data within an Earwurm?
    • Otherwise, we will sometimes have a "delay" before audio begins playing for the first time.
    • It would also be nice if a Stack had a cached boolean getter.
  10. Is it reasonable to cache the decoded AudioBuffer by storing it on the Stack instance?
    • We could introduce this private property: #audioBuffer: AudioBuffer | null;
    • And then add an option to the config for cacheAudioBuffer?: boolean;
    • This would have to be flagged as potentially "dangerous", as I imagine caching too many buffers creates considerable overhead.
  11. Do we need to re-check the #queue state when a sound has ended?
    • This would be done within Stack > #handleSoundEnded.
    • A sound goes into a stopping state first, which will trigger #setStateFromQueue. There shouldn't be any reason for the ended event to fail… but if so, then we will need to re-check.
  12. Find out if statechange event listeners automatically get removed upon calling .close() / state change to closed on a AudioContext.
    • If so, then I can remove the .close().then() chain.
  13. Consider adding a readonly parent to both Stack and Sound.
    • The value of each would be this from either the Manager or the Stack.
    • Example: This would allow me to know what Stack manages a particular Sound.
  14. Consider setting Stack > state to closed on teardown.
    • This way, there is a better mechanism to removing a Stack after listening to a statechange.

Additional

Maybe there is some value in passing an array of GainNode/AudioNode for destination? If so, that might look something like:

export type AudioOutputs = [...GainNode[], AudioNode];

constructor(readonly outputs: AudioOutputs);

let lastConnection: GainNode | AudioNode = this.#gainNode;

outputs.forEach((node) => {
  lastConnection = lastConnection.connect(node);
});