3
0
Fork 0
mirror of https://gitea.com/actions/setup-python.git synced 2024-11-25 19:49:35 +01:00
setup-python/node_modules/throat
Danny McCormick 39c08a0eaa Initial pass
2019-06-26 21:12:00 -04:00
..
index.d.ts Initial pass 2019-06-26 21:12:00 -04:00
index.js Initial pass 2019-06-26 21:12:00 -04:00
index.js.flow Initial pass 2019-06-26 21:12:00 -04:00
LICENSE Initial pass 2019-06-26 21:12:00 -04:00
package.json Initial pass 2019-06-26 21:12:00 -04:00
README.md Initial pass 2019-06-26 21:12:00 -04:00

throat

Throttle the parallelism of an asynchronous, promise returning, function / functions. This has special utility when you set the concurrency to 1. That way you get a mutually exclusive lock.

Build Status Coverage Status Dependency Status NPM version Greenkeeper badge

Sauce Test Status

Installation

npm install throat

API

throat(concurrency)

This returns a function that acts a bit like a lock (exactly as a lock if concurrency is 1).

Example, only 2 of the following functions will execute at any one time:

const throat = require('throat')(2);
// alternatively provide your own promise implementation
const throat = require('throat')(require('promise'))(2);
const promise = Promise.resolve();

const resA = throat(() => /* async stuff... */ promise);
const resB = throat(() => /* async stuff... */ promise);
const resC = throat(() => /* async stuff... */ promise);
const resD = throat(() => /* async stuff... */ promise);
const resE = throat(() => /* async stuff... */ promise);

throat(concurrency, worker)

This returns a function that is an exact copy of worker except that it will only execute up to concurrency times in parallel before further requests are queued:

const throat = require('throat');
// alternatively provide your own promise implementation
const throat = require('throat')(require('promise'));

const input = ['fileA.txt', 'fileB.txt', 'fileC.txt', 'fileD.txt'];
const data = Promise.all(input.map(throat(2, fileName => readFile(fileName))));

Only 2 files will be read at a time, sometimes limiting parallelism in this way can improve scalability.

License

MIT