This is a message-passing abstraction layer implemented by Typescript. Its purpose is to encapsulate the details of messaging, allowing js to construct classes that exist in different contexts and call methods in different contexts in an idiomatic way.
- Encapsulation: Encapsulates communication details and provides consistent upper-layer APIs
- Isolation: Data is securely isolated in different channel-ids.
- API: Idiomatic API, remote calls are as simple as ordinary asynchronous calls, support callback functions, and exception handling is consistent with ordinary asynchronous methods
- Parallel: support decomposing tasks into multiple target contexts for parallel execution
- Extension: support custom communicator to for more features
npm i -s @vgerbot/channel
Then install @vgerbot/channel-transformer
, this is to simplify things and make channel APIs more convenient to use
npm i -D @vgerbot/channel-transformer
For more information about the usage of @vgerbot/channel-transformer
please refer https://github.com/vgerbot-libraries/tschannel/blob/master/packages/transformer/README.md
api.ts
export interface SpellChecker {
saveToDictionary(word: string): void;
setCaseSensitive(caseSensitive: boolean): void;
check(sentence: string): boolean;
}
worker.ts
import { channel } from '@vgerbot/channel';
import { SpellChecker } form './api';
const chnl = channel('worker-channel')
.connectToMainThread()
.create();
chnl.def_method(function performCPUIntensiveCalculation() {
return 'Result!';
});
chnl.def_class(class DefaultSpellCheckerImpl implements SpellChecker {
saveToDictionary(word: string) {}
setCaseSensitive(caseSensitive: boolean) {}
check(sentence) {
return true;
}
})
client.ts
import { channel } from '@vgerbot/channel';
import { SpellChecker } form './api';
const chnl = channel('worker-channel')
.connectToWorker('./worker.js') // This is the path to the compiled js file for worker.ts
.create();
const performCPUIntensiveCalculation = chnl.get_method<() => string>('performCPUIntensiveCalculation');
performCPUIntensiveCalculation().then(console.log) // Console Output: "Result!"
const DefaultSpellCheckerImpl = chnl.get_class<SpellChecker>('DefaultSpellCheckerImpl');
// DefaultSpellCheckerImpl is defined as follows
/*
class DefaultSpellCheckerImpl {
saveToDictionary(word: string): Promise<void> {
// REMOVE METHOD
}
setCaseSensitive(caseSensitive: boolean): Promise<void> {
// REMOVE METHOD
}
check(sentence: string): Promise<boolean> {
// REMOVE METHOD
}
}
*/
// You can construct an instance as you would a normal object, and then call the interface methods, except that they are all asynchronous, i.e. the return value is a Promise
const spellChecker = new DefaultSpellCheckerImpl();
spellChecker.saveToDictionary('halo');
spellChecker.setCaseSensitive(false);
spellChecker.check('Halo world!').then(console.log); // Console Output: true
spellChecker.__destroy__(); // Since the remote instance cannot be automatically cleared by the GC, it must be destroyed manually.
For more examples, please refer to examples and unit tests.
Like the postMessage API, it supports all types that can be cloned using the structured clone algorithm. For more detailed description, please refer toThe structured clone algorithm In addition to supporting the parameter types of postMessage, remote objects and callback functions are also supported, but these two types cannot be nested in other objects.
The @vgerbot/channel
library is released under the terms of the .