I'm going through the "Functional Javascript Workshop" through Nodeschool. One of the exercises is titled "Blocking Event Loop" and I am having difficulty understanding it. With past exercises, I have made sure to really try to understand the solution so that if I were to have to redo the problem, I would understand how to solve it (as opposed to hacking away at it the first time). But this concept is really challenging me.
Modify the recursive repeat function provided in the boilerplate, such
that it does not block the event loop (i.e. Timers and IO handlers can
fire). This necessarily requires repeat to be asynchronous.
A timeout is queued to fire after 100 milliseconds, which will print the
results of the test and exit the process. repeat should release
control of the event loop to allow the timeout to interrupt before all the operations complete.
Try to perform as many operations as you can before the timeout fires!
Boilerplate
function repeat(operation, num) {
// modify this so it can be interrupted
if (num <= 0) return
operation()
return repeat(operation, --num)
}
module.exports = repeat
Solution
function repeat(operation, num) {
if (num <= 0) return
operation()
// release control every 10 or so
// iterations.
// 10 is arbitrary.
if (num % 10 === 0) {
setTimeout(function() {
repeat(operation, --num)
})
} else {
repeat(operation, --num)
}
}
module.exports = repeat
I have sought to understand setTimeout better and I somewhat understand how it works. But I do not understand some of the language and concepts in the question itself:
Modify the recursive repeat function provided in the boilerplate,
such that it does not block the event loop (i.e. Timers and IO handlers can fire). This necessarily requires repeat to be
asynchronous.
I don't understand how the solution, which makes every 10th repeat async, prevents repeat from blocking the event loop. I'm assuming the event loop and event queue are like the same things? For instance clicks get placed on the event queue while the javascript code is running synchronously through its code until the execution stack is empty and then the queue gets looked at. How does making every 10th repeat async prevent the queue from being blocked -- my understanding of "being blocked" is that items in the queue are blocked until the stack is empty, at which point javascript looks at what is in the queue. So lets say at repeat 10, 20, 30 etc, this code makes those repeats async, but doesn't it still continue through all the nums until it hits zero, at which point the stack is now empty, before javascript looks at the queue? So how does this solution resolve that blocking like the question asks?
Next, the question refers to "releasing control". No clue what that means. What is control and how is it released and what is it released to?
Immediately after that the question says "allow the timeout to interrupt before all the operations complete." Does interruption mean that basically every 10th repeat is interrupted (not allowed to complete until the synchronous code ends)?
Finally, the challenge at the end to try and perform as many operations as one can before the timeout fires... I also don't see how that applies to the solution. The setTimeout doesn't even seem to have a set duration.
The message queue (you refer to it as the event queue) is a list of messages to be processed
The event loop processes these messages one by one to completion
The blocking version of the repeat function you posted would be
function repeat(operation, num) {
if (num <= 0) return
operation()
repeat(operation, --num)
}
You can see that this recursively calls itself until num <= 0, so this wont complete until num <= 0, therefore no new messages will be processed by the event loop until this finishes
Contrast this with the non blocking version
function repeat(operation, num) {
if (num <= 0) return
operation()
// release control every 10 or so
// iterations.
// 10 is arbitrary.
if (num % 10 === 0) {
setTimeout(function() {
repeat(operation, --num)
})
} else {
repeat(operation, --num)
}
}
every 10 iterations, rather than recursively calling repeat, this function places (by virtue of the setTimeout) the next call to repeat to the end of the message queue and does nothing else (therefore is completed)
this allows the event loop to process any and all messages that were placed on the message queue during the 10 iterations of repeat before the setTimeout call, and only once those messages have been processed to completion will the callback in setTimeout be executed, which will be the next iteration of the repeat function
I have the following code that involves an IIFE (immediately invoking function expression) within a for loop. The IIFE function is clearly getting the right parameter as the printout is as expected. But I don't understand what the interval is doing. As far as I can tell the interval for the 1st iteration should be 1 sec, then 2 sec for the 2nd iteration, etc. etc.
for (var i = 1; i <= 5; i++) {
(function(i){
setTimeout(function timer(){
console.log(i);
}, i*1000);
})(i);
}
I see i printed out as 1..5 in 1 second intervals. This is confusing for me, as I was expecting the interval to increase with each iteration.
With the following code:
for (var i = 1; i <= 5; i++) {
(function(i){
setTimeout(function timer(){
console.log(i);
}, 1000);
})(i);
}
I see all the values for i printed at once after a 1 second interval.
What is happening within the setTimeout function that is making it work in the way observed?
The for loop runs to completion. It doesn't wait for the setTimeout() call to fire. So, when the for loop is running, it schedules setTimeout() calls for 1000, 2000, 3000, 4000, 5000 ms from now. Thus, you see them all fire one second apart.
What is happening within the setTimeout function that is making it
work in the way observed?
setTimeout() is a non-blocking, asynchronous operation. It is scheduled to run in the future, but the other code around it continues to run. Then, sometime in the future the timer fires and the callback is called.
So, your code is essentially this:
setTimeout(fn, 1000);
setTimeout(fn, 2000);
setTimeout(fn, 3000);
setTimeout(fn, 4000);
setTimeout(fn, 5000);
It runs all the setTimeout() calls one after the other in a non-blocking fashion. Each setTimeout() function call just registers a callback and a time with the underlying system and then the next line of code continues to execute. Then, sometime later when the right amount of time has passed, the callback gets called. So, hopefully you can see that all the timers are all scheduled to measure their time starting from now. Thus, you see them fire each 1 second apart.
If you want an analogy for an asynchronous operation, imagine you walk around your house and you set five alarm clocks to start ringing at a time in the future. You set one to ring an hour from now, one for two hours from now, one for three hours from now, etc... Then, you go about your other business, make lunch, clean the kitchen, mow the grass, etc... and one hour from when you set all the clocks, the first alarm starts ringing. One hour after that, the next one starting ringing, etc... You get to keep doing things before and after they ring. You don't have to sit and wait for a clock to ring before doing anything else. That's what Javascript asynchronous timer operations are like. You set them and then keep running other code and they fire at their scheduled time in the future.
Try this :
function recursiveTimeout(i) {
setTimeout(function timer(){
console.log(i);
if(i<5) {
recursiveTimeout(i+1);
}
}, 1000 * i);
}
recursiveTimeout(1);
That way the next setTimeout is depending of the end of parent timer.
As far as I can tell, these two pieces of javascript behave the same way:
Option A:
function myTimeoutFunction()
{
doStuff();
setTimeout(myTimeoutFunction, 1000);
}
myTimeoutFunction();
Option B:
function myTimeoutFunction()
{
doStuff();
}
myTimeoutFunction();
setInterval(myTimeoutFunction, 1000);
Is there any difference between using setTimeout and setInterval?
They essentially try to do the same thing, but the setInterval approach will be more accurate than the setTimeout approach, since setTimeout waits 1000ms, runs the function and then sets another timeout. So the wait period is actually a bit more than 1000ms (or a lot more if your function takes a long time to execute).
Although one might think that setInterval will execute exactly every 1000ms, it is important to note that setInterval will also delay, since JavaScript isn't a multi-threaded language, which means that - if there are other parts of the script running - the interval will have to wait for that to finish.
In this Fiddle, you can clearly see that the timeout will fall behind, while the interval is almost all the time at almost 1 call/second (which the script is trying to do). If you change the speed variable at the top to something small like 20 (meaning it will try to run 50 times per second), the interval will never quite reach an average of 50 iterations per second.
The delay is almost always negligible, but if you're programming something really precise, you should go for a self-adjusting timer (which essentially is a timeout-based timer that constantly adjusts itself for the delay it's created)
Is there any difference?
Yes. A Timeout executes a certain amount of time after setTimeout() is called; an Interval executes a certain amount of time after the previous interval fired.
You will notice the difference if your doStuff() function takes a while to execute. For example, if we represent a call to setTimeout/setInterval with ., a firing of the timeout/interval with * and JavaScript code execution with [-----], the timelines look like:
Timeout:
. * . * . * . * .
[--] [--] [--] [--]
Interval:
. * * * * * *
[--] [--] [--] [--] [--] [--]
The next complication is if an interval fires whilst JavaScript is already busy doing something (such as handling a previous interval). In this case, the interval is remembered, and happens as soon as the previous handler finishes and returns control to the browser. So for example for a doStuff() process that is sometimes short ([-]) and sometimes long ([-----]):
. * * • * • * *
[-] [-----][-][-----][-][-] [-]
• represents an interval firing that couldn't execute its code straight away, and was made pending instead.
So intervals try to ‘catch up’ to get back on schedule. But, they don't queue one on top of each other: there can only ever be one execution pending per interval. (If they all queued up, the browser would be left with an ever-expanding list of outstanding executions!)
. * • • x • • x
[------][------][------][------]
x represents an interval firing that couldn't execute or be made pending, so instead was discarded.
If your doStuff() function habitually takes longer to execute than the interval that is set for it, the browser will eat 100% CPU trying to service it, and may become less responsive.
Which do you use and why?
Chained-Timeout gives a guaranteed slot of free time to the browser; Interval tries to ensure the function it is running executes as close as possible to its scheduled times, at the expense of browser UI availability.
I would consider an interval for one-off animations I wanted to be as smooth as possible, whilst chained timeouts are more polite for ongoing animations that would take place all the time whilst the page is loaded. For less demanding uses (such as a trivial updater firing every 30 seconds or something), you can safely use either.
In terms of browser compatibility, setTimeout predates setInterval, but all browsers you will meet today support both. The last straggler for many years was IE Mobile in WinMo <6.5, but hopefully that too is now behind us.
setInterval()
setInterval() is a time interval based code execution method that has the native ability to repeatedly run a specified script when the interval is reached. It should not be nested into its callback function by the script author to make it loop, since it loops by default. It will keep firing at the interval unless you call clearInterval().
If you want to loop code for animations or on a clock tick, then use setInterval().
function doStuff() {
alert("run your code here when time interval is reached");
}
var myTimer = setInterval(doStuff, 5000);
setTimeout()
setTimeout() is a time based code execution method that will execute a script only one time when the interval is reached. It will not repeat again unless you gear it to loop the script by nesting the setTimeout() object inside of the function it calls to run. If geared to loop, it will keep firing at the interval unless you call clearTimeout().
function doStuff() {
alert("run your code here when time interval is reached");
}
var myTimer = setTimeout(doStuff, 5000);
If you want something to happen one time after a specified period of time, then use setTimeout(). That is because it only executes one time when the specified interval is reached.
The setInterval makes it easier to cancel future execution of your code. If you use setTimeout, you must keep track of the timer id in case you wish to cancel it later on.
var timerId = null;
function myTimeoutFunction()
{
doStuff();
timerId = setTimeout(myTimeoutFunction, 1000);
}
myTimeoutFunction();
// later on...
clearTimeout(timerId);
versus
function myTimeoutFunction()
{
doStuff();
}
myTimeoutFunction();
var timerId = setInterval(myTimeoutFunction, 1000);
// later on...
clearInterval(timerId);
I find the setTimeout method easier to use if you want to cancel the timeout:
function myTimeoutFunction() {
doStuff();
if (stillrunning) {
setTimeout(myTimeoutFunction, 1000);
}
}
myTimeoutFunction();
Also, if something would go wrong in the function it will just stop repeating at the first time error, instead of repeating the error every second.
The very difference is in their purposes.
setInterval()
-> executes a function, over and over again, at specified time intervals
setTimeout()
-> executes a function, once, after waiting a specified number of milliseconds
It's as simple as that
More elaborate details here http://javascript.info/tutorial/settimeout-setinterval
When you run some function inside setInterval, which works more time than timeout-> the browser will be stuck.
- E.g., doStuff() takes 1500 sec. to be execute and you do: setInterval(doStuff, 1000);
1) Browser run doStuff() which takes 1.5 sec. to be executed;
2) After ~1 second it tries to run doStuff() again. But previous doStuff() is still executed-> so browser adds this run to the queue (to run after first is done).
3,4,..) The same adding to the queue of execution for next iterations, but doStuff() from previous are still in progress...
As the result- the browser is stuck.
To prevent this behavior, the best way is to run setTimeout inside setTimeout to emulate setInterval.
To correct timeouts between setTimeout calls, you can use self-correcting alternative to JavaScript's setInterval technique.
Your code will have different execution intevals, and in some projects, such as online games it's not acceptable. First, what should you do, to make your code work with same intevals, you should change "myTimeoutFunction" to this:
function myTimeoutFunction()
{
setTimeout(myTimeoutFunction, 1000);
doStuff();
}
myTimeoutFunction()
After this change, it will be equal to
function myTimeoutFunction()
{
doStuff();
}
myTimeoutFunction();
setInterval(myTimeoutFunction, 1000);
But, you will still have not stable result, because JS is single-threaded. For now, if JS thread will be busy with something, it will not be able to execute your callback function, and execution will be postponed for 2-3 msec. Is you have 60 executions per second, and each time you have random 1-3 sec delay, it will be absolutely not acceptable (after one minute it will be around 7200 msec delay), and I can advice to use something like this:
function Timer(clb, timeout) {
this.clb = clb;
this.timeout = timeout;
this.stopTimeout = null;
this.precision = -1;
}
Timer.prototype.start = function() {
var me = this;
var now = new Date();
if(me.precision === -1) {
me.precision = now.getTime();
}
me.stopTimeout = setTimeout(function(){
me.start()
}, me.precision - now.getTime() + me.timeout);
me.precision += me.timeout;
me.clb();
};
Timer.prototype.stop = function() {
clearTimeout(this.stopTimeout);
this.precision = -1;
};
function myTimeoutFunction()
{
doStuff();
}
var timer = new Timer(myTimeoutFunction, 1000);
timer.start();
This code will guarantee stable execution period. Even thread will be busy, and your code will be executed after 1005 mseconds, next time it will have timeout for 995 msec, and result will be stable.
I use setTimeout.
Apparently the difference is setTimeout calls the method once, setInterval calls it repeatdly.
Here is a good article explaining the difference: Tutorial: JavaScript timers with setTimeout and setInterval
I've made simple test of setInterval(func, milisec), because I was curious what happens when function time consumption is greater than interval duration.
setInterval will generally schedule next iteration just after the start of the previous iteration, unless the function is still ongoing. If so, setInterval will wait, till the function ends. As soon as it happens, the function is immediately fired again - there is no waiting for next iteration according to schedule (as it would be under conditions without time exceeded function). There is also no situation with parallel iterations running.
I've tested this on Chrome v23. I hope it is deterministic implementation across all modern browsers.
window.setInterval(function(start) {
console.log('fired: ' + (new Date().getTime() - start));
wait();
}, 1000, new Date().getTime());
Console output:
fired: 1000 + ~2500 ajax call -.
fired: 3522 <------------------'
fired: 6032
fired: 8540
fired: 11048
The wait function is just a thread blocking helper - synchronous ajax call which takes exactly 2500 milliseconds of processing at the server side:
function wait() {
$.ajax({
url: "...",
async: false
});
}
Both setInterval and setTimeout return a timer id that you can use to cancel the execution, that is, before the timeouts are triggered. To cancel you call either clearInterval or clearTimeout like this:
var timeoutId = setTimeout(someFunction, 1000);
clearTimeout(timeoutId);
var intervalId = setInterval(someFunction, 1000),
clearInterval(intervalId);
Also, the timeouts are automatically cancelled when you leave the page or close the browser window.
To look at it a bit differently: setInterval ensures that a code is run at every given interval (i.e. 1000ms, or how much you specify) while setTimeout sets the time that it 'waits until' it runs the code. And since it takes extra milliseconds to run the code, it adds up to 1000ms and thus, setTimeout runs again at inexact times (over 1000ms).
For example, timers/countdowns are not done with setTimeout, they are done with setInterval, to ensure it does not delay and the code runs at the exact given interval.
You can validate bobince answer by yourself when you run the following javascript or check this JSFiddle
<div id="timeout"></div>
<div id="interval"></div>
var timeout = 0;
var interval = 0;
function doTimeout(){
$('#timeout').html(timeout);
timeout++;
setTimeout(doTimeout, 1);
}
function doInterval(){
$('#interval').html(interval);
interval++;
}
$(function(){
doTimeout();
doInterval();
setInterval(doInterval, 1);
});
Well, setTimeout is better in one situation, as I have just learned. I always use setInterval, which i have left to run in the background for more than half an hour. When i switched back to that tab, the slideshow (on which the code was used) was changing very rapidly, instead of every 5 seconds that it should have. It does in fact happen again as i test it more and whether it's the browser's fault or not isn't important, because with setTimeout that situation is completely impossible.
The difference is obvious in console:
Just adding onto what has already been said but the setTimeout version of the code will also reach the Maximum call stack size which will stop it from functioning. Since there is no base case for the recursive function to stop at so you can't have it run forever.
If you set the interval in setInterval too short, it may fire before the previous call to the function has been completed. I ran into this problem with a recent browser (Firefox 78). It resulted in the garbage collection not being able to free memory fast enough and built up a huge memory leak.
Using setTimeout(function, 500); gave the garbage collection enough time to clean up and keep the memory stable over time.
Serg Hospodarets mentioned the problem in his answer and I fully agree with his remarks, but he didn't include the memory leak/garbage collection-problem. I experienced some freezing, too, but the memory usage ran up to 4 GB in no time for some minuscule task, which was the real bummer for me. Thus, I think this answer is still beneficial to others in my situation. I would have put it in a comment, but lack the reputation to do so. I hope you don't mind.
The reason why Option A and Option B seem like they work the same is mostly because the places of the setInterval and the setTimeout functions.
function myTimeoutFunction()
{
doStuff();
setTimeout(myTimeoutFunction, 1000);
}
myTimeoutFunction();
This one is a recursive function, and if doStuff is very complex, setTimeout has to keep track of all calls of the setTimout plus the current doStuff, which makes it become slower and s l o w e r.
function myTimeoutFunction()
{
doStuff();
}
myTimeoutFunction();
setInterval(myTimeoutFunction, 1000);
On the other hand, the setInterval only has to keep track of the last setInterval and the current doStuff, making it staying at a constant speed.
So which one should you use?
From the above, you should probably be able to conclude that the better one is setInterval.
The important point to consider is the performance.
The only way to run a function periodically using setTimeout is to call it recursively with the target function, and when you check it, it appears that it works asynchronously put when you see the call stack you will find it keep growing by the time. In fact, it is sensible. Since Javascript does not support multi-threading, it is impossible to finish calling the parent function before finishing the child function, therefor, the stack will keep growing as long as there is recursive calling.
Whilst, with setInterval we don't need to call the target function recursively since it has a logic that runs it periodically as a loop. So, this keeps the call stack clean.
You can watch the call stack using developer's tools in your browser and you will notice the difference.
The difference will be clear when using small interval for a long period of time.
I think SetInterval and SetTimeout are different. SetInterval executes the block according to the time set while, SetTimeout executes the block of code once.
Try these set of codes after the timeout countdown seconds:
setInterval(function(e){
alert('Ugbana Kelvin');
}, 2000);
and then try
setTimeout(function(e){
alert('Ugbana Kelvin');
}, 2000);
You can see the differences for yourself.