three-wtm

WorkerTaskManager for three.js

Overview

The WorkerTaskManager allows to register tasks expressed by an initialization, and an execution function with an optional comRounting function to be run in a web worker. It creates one to a maximum number of workers that can be used for execution. Multiple execution requests can be handled in parallel of the main task, If all workers are currently occupied the requested are enqueued, and the returned promise is fulfilled once a worker becomes available again.

Features

• WorkerTaskManager supports standard workers:
  • Code for standard workers can be created from module code on main or just provided as file obeying the standard worker.
  • exec and init functions can be declared in module and then be packaged in standard worker if needed. I used this to define the worker code once and support both code paths
  • Loading of dependencies for standard workers is available
  • Main-Execution fallback is available
• WorkerTaskManager supports module workers:
  • Dependencies are declared regularly and therefore no extra functionality is needed
• Code is written in ES6+ notation and uses ES6+ features and class notation is used.
• WorkerTaskManager has an execution queue allowing to queue more task than can be currently handled OBJLoader2
• WorkerTaskManager is used by wwobjloader2 and it is potentially added to three.js itself.
• Example shows webgl_loader_workertaskmanager.html how an existing OBJLoader can be wrapped in a worker.
• Additional utility functions are available allowing to transfer BufferGeometry, Material (meta-information) and Mesh bi-driectionally between Main and workers. All "transferables" shall be treated as such.

Getting Started

There exist three possibilities:

• Press the Gitpod button above and start coding and using the examples directly in the browser
• Checkout the repository and use docker-compose up -d to spin up local snowpack dev server
• Checkout the repository and run npm install and then npm run dev to spin up local snowpack dev server

Whatever environment you choose to start snowpack is used to serve the code and the examples using it. With this setup you are able to change the code and examples without invoking an additional bundler. Snowpack ensures all imported npm modules are available if previously installed in local environment (see npm install).

If you run snowpack locally you require a nodejs and npm. The Gitpod and local docker environment ensure all prerequisites are fulfilled.

In any environment the dev server is reachable on port 8080.

Main Branches

Main development takes place on branch main. The stable branch contains the release versions.

Docs

Run npm run doc to create the documentation in directory build/docs.

Examples

This gives you an idead how you can define a standard worker inline and the register it with WorkerTaskManager, init and execute it:

const init = function ( context, id, config ) {
  /* init code ... */
  context.postMessage( { cmd: "init", id: id } );
};

const execute = function ( context, id, config ) {
  /* execution code ... */
  context.postMessage( { cmd: "execComplete", data: {} } );
}

const buildDependencies = function () {
  return [
    { url: '/node_modules/three/build/three.min.js' },
    { code: 'const hello = "Hello World!";' }
  ]
}

const taskName = 'InlineWorker';
this.workerTaskManager.registerTaskType(taskName, init, execute, null, false, buildDependencies());
await this.workerTaskManager.initTaskType(taskName, { name: taskName })
    .then( () => this.workerTaskManager.enqueueForExecution( taskName, {}, null))
    .then( data => processData(data) )
    .catch( e => console.error( e ) );

This is the same block required for a module Worker. The worker code resides in an extra file an contains all imports and exports required:

const taskNameModule = 'ModuleWorker';
this.workerTaskManager.registerTaskTypeModule('myModuleTask', '/examples/worker/helloWorldWorker.js');
await this.workerTaskManager.initTaskType(taskNameModule, {})
    .then( () => this.workerTaskManager.enqueueForExecution( taskName, {}, null))
    .then( data => processData(data) )
    .catch( e => console.error( e ) );
    

The examples above are fully specified and used in wtm_helloworld.html.

Bundling

Rollup is used to create the libraries that are contained in the deployed npm package. Just run npm run build

Local library test environment

Rollup is also used to create local test environments where the library and uglified library are used by the examples. Use npm run prerelease to build the libraries and the test environment. Inside both build/verify or build/verifymin you find a npm package.json and snowpack configuration that allows to serve everything. Run npm run dev. The example then use the packages library.

Happy coding!

Kai