Best performance practice while using modules

Reply to Best performance practice while using modules on Invalid Date

Mototroller — Invalid Date

There is a question about what the best performance practice to use modules.

In the case we use module to determine creep's role behavior, for example, which variant will be more CPU-effective:

Module as function (from documentation):

/// harvester.js
module.exports = function(creep) {
    /// Some private vars
    /// Some supporting functions
    return {
        work : function() { /* JOB */ }
    }
}

/// main.js
var harvester = require('harvester');
for(var i in Game.creeps) {
    harvester(Game.creeps[i]).work();
}

... or module as a result of function call:

/// harvester.js
module.exports = (function() {
    /// Some private vars
    /// Some supporting functions
    return {
        work : function(creep) { /* JOB */ }
    }
}());

/// main.js
var harvester = require('harvester');
for(var i in Game.creeps) {
    harvester.work(Game.creeps[i]);
}

... or any other way?

Have anybody thought about this and mb did any cpu-usage measures?

Reply to Best performance practice while using modules on Invalid Date

neomatrix — Invalid Date

In terms of "CPU performance": each require call takes 0.6 cpu on the average, no matter how you use it's result further.
(that is according to my own measures)

So, if you're interested in a high performance, you may think about minimising the number of modules (the number of require calls in your script) rather then it's further usage.

About your initial question: I don't see any evident difference in it (due to measures I did), but there may be other opinions.

Reply to Best performance practice while using modules on Invalid Date

Mototroller — Invalid Date

Хорошо, спасибо.

P.S. Увидел в соседней ветке подробную расписку по замерам

Reply to Best performance practice while using modules on Invalid Date

OkkeHendriks — Invalid Date

Interesting, so there is a direct relationship between maintainability and cost (assuming code is more maintainable when divided into modules). For you to decide if its worth it!

Reply to Best performance practice while using modules on Invalid Date

Hatyr — Invalid Date

I've found that each require call takes an amount of time correlated to how much the file contains... Following neomatrix's testing I created a bundle containing all my code then measured it's require time in main.js... 23 ms.