Fast microtask queue for all platforms, equivalent rawAsap (based on the ideas and source of rawAsap), but a little faster.
var nextTask = require('next-task');
nextTask(function() {
/* this === undefined, arguments.length === 0 */
console.log('Run this async, but "as soon as possible".');
});
console.log('This run sync.');
/** Log:
* -> This run sync.
* -> Run this async, but "as soon as possible".
*/
/** Variant with context: */
var task = {
data: ...,
call: function() {/* this === task */}
};
nextTask(task);
About rawAsap and microtasks: rawAsap. If you need queue of animation tasks, use raf instead, for synchronize with rendering loop. If you need to perform a long (macrotask) queue of heavy tasks, use setImmediate to give the browser the ability to handle current events. Note that, like rawAsap, next-task does not catch the errors (to work as soon as possible).
/**
* If a task does throw an error, with rawAsap you need
* call requestFlush (after error):
*/
rawAsap.requestFlush();
/**
* With nextTask just call it without arguments
*(or with null or undefined), also after error:
*/
nextTask();
next-task does not support domains for Node.js (rasAsap does).
next-task uses native Promise, if it is available (only native, and ignores any polyfills). More information: Consider using Promise.prototype.then.
Property 'use' points to the technology used:
/** In the order of attempts to use: */
nextTask.use === 'setImmediate' || /* only Node.js */
'Promise' || /* ES6 native promise, if available */
'MutationObserver' || /* modern browsers */
'setTimeout' /* all other platforms */
Method 'setCapacity' limited the memory usage (more information: function to change rawAsap.capacity value must be added):
nextTask.setCapacity(1024); /* return 1024 */
nextTask.setCapacity(); /* return 1024 */
nextTask.setCapacity({}); /* return 1024 */
nextTask.setCapacity(100); /* return 100 */
Install all the packages from devDependencies in ./node_modules and run build:
$ npm install
$ npm run build
Then you will be able to perform tests and benchmark.
$ npm run benchmark:node
$ npm run benchmark:browser
This is benchmark from asap/benchmarks, in which different ways queuing added for comparison. The results are not very stable and not fully explained; for example, a typical result in Node.js:
asap x 5,584 ops/sec ±3.90% (56 runs sampled)
rawAsap x 39,700 ops/sec ±7.41% (56 runs sampled)
nextTask x 41,186 ops/sec ±7.55% (57 runs sampled)
nextTick x 11,711 ops/sec ±5.81% (58 runs sampled)
nextTick[] x 23,823 ops/sec ±3.49% (59 runs sampled)
setImmediate x 5,860 ops/sec ±4.92% (59 runs sampled)
setImmediate[] x 14,182 ops/sec ±3.91% (59 runs sampled)
Promise x 811 ops/sec ±3.43% (58 runs sampled)
Promise[] x 11,993 ops/sec ±1.89% (55 runs sampled)
setTimeout x 612 ops/sec ±1.95% (59 runs sampled)
setTimeout[] x 768 ops/sec ±1.32% (58 runs sampled)
The results of benchmark in DOM even more dependent on the browser. For example, in modern Chrome:
asap x 6,770 ops/sec ±4.53% (42 runs sampled)
rawAsap x 27,173 ops/sec ±7.02% (42 runs sampled)
nextTask x 32,403 ops/sec ±3.36% (46 runs sampled)
Promise x 891 ops/sec ±3.33% (44 runs sampled)
Promise[] x 10,534 ops/sec ±2.21% (42 runs sampled)
MutationObserver[] x 4,792 ops/sec ±5.05% (41 runs sampled)
setTimeout x 115 ops/sec ±1.56% (47 runs sampled)
setTimeout[] x 226 ops/sec ±1.85% (50 runs sampled)
There "Promise" is run each task by Promise and "Promise[]" means the use of the task queue, so that the whole queue is run with one call Promise; and similarly for the other methods.
$ npm run test:node
$ npm run test:browser