Poel
Create a pool of cluster workers.
Usage
const poel = require('poel'); (async () => { const pool = await poel({ getPid, sayHi, }); for (let i = 0; i < 10; i++) { // On an 8 core system, this should print 8 unique pids // and then start then start repeating pids. console.log(await pool.getPid()) } console.log(await pool.sayHi('World')) await pool.$.shutdown(); })(); function getPid() { return process.pid; } function sayHi(person) { return new Promise((resolve) => { setTimeout(() => resolve(`Hi ${person}!`), 200); }); }
Explanation
poel takes an object of functions which it then returns as
a pool of workers with those same functions. In the example
above, it was passed the getId and sayHi functions. The
pool object returned also has a getId and sayHi
method. Calling any method will cycle through the workers, so
in the example above it will print as many unique numbers as
there CPU cores on the system. If we called sayHi inside
the loop, then the output from the getPid call would never
include the pids of the odd numbered workers.
Gotchas
poel uses cluster
mode under the hood. The main thing to keep in mind is that
you must await (or Promise.resolve()) the call to poel
before the main logic of your script. The call to poel will
only resolve for the master process, which most of the time
is really all you want.
Another thing to watch out for is that value passed into and
back out of functions go through JSON.stringify and
JSON.parse, so passing regexes and functions won't work.
jsan is one solution to
pass complex objects around.
The example above takes advantage of
function hoisting
so as to be able to have the main logic near the top of the
file. If the functions were instead declared as
const getPid = () => process.pid then you would need to
have that line above the call to poel.
/pōel/ פועל. po as in Edgar Allan Poe, el as in the 12th
letter of the alphabet (L).
Poel means worker in Hebrew.