Bačen za petlju: razumijevanje petlji i vremenskih ograničenja u JavaScript-u

JavaScript često samo radi. A budući da je napisan u ljudskom čitljivoj sintaksi, neke se stvari čine intuitivnima. Ali lako je zanemariti ono što se događa na dubljoj razini. Na kraju, međutim, taj nerazumijevanje rezultira nemogućnošću rješavanja problema.

Intuicija je sposobnost da se nešto odmah shvati, bez potrebe za svjesnim rasuđivanjem. - Google

Provodim prilično vremena pokušavajući riješiti dvodimenzionalne probleme, a nešto veći dio pokušavajući riješiti trodimenzionalne.

Iako uživam vježbati kodiranje u slobodno vrijeme, danju sam kontrolor zračnog prometa. Problemi s kojima se susrećemo kao kontrolori zračnog prometa ne razlikuju se od bilo kojeg drugog posla. Postoje rutinski problemi s rutinskim rješenjima i jedinstveni problemi s jedinstvenim rješenjima. Kroz dublje razumijevanje možemo riješiti jedinstvene.

Izvana gledajući kontrolu zračnog prometa, može se činiti da je sve jedinstveni problem - da postoji svojstvena potrebna vještina za obavljanje posla. Međutim, iako određene sklonosti mogu olakšati učenje bilo koje vještine, u konačnici je iskustvo ono što rješavanje problema tjera na podsvjesnu razinu. Rezultat je intuicija.

Intuicija slijedi nakon promatranja. Dovoljno puta uočite jedinstveni problem i on i njegovo rješenje postaju rutina. Primjećuje dosljednost u svakoj situaciji u kojoj počinjemo razvijati osjećaj što bi se trebalo dogoditi sljedeće.

Intuicija , međutim, ne zahtijeva duboko razumijevanje. Često možemo ukazati na točno rješenje, a da ne možemo artikulirati kako i zašto ono djeluje. Međutim, ponekad biramo rješenja koja se čine intuitivnima, a zapravo njima upravlja nepoznati skup pravila.

Što daje ovaj kod?

for(var i = 1; i < 6; i++) { setTimeout(function() { console.log(i); },1000); } console.log('The loop is done!');

Odvojite malo vremena da razmislite o tome što će dati ovaj kod. Počet ćemo graditi temelje za odgovor na ovo, a na to ćemo se vratiti kasnije.

JavaScript je jezični dijalekt.

Odrastao sam na sjeveroistoku SAD-a. Iako govorim engleski, moj govor nedvojbeno sadrži regionalnu raznolikost. Ova se sorta naziva dijalektom . Moj određeni dijalekt je provedba (ili verzija) standarda na engleskom jeziku.

Možda se čini da bi standardi rađali dijalekte, ali dijalekt je taj koji pokreće potrebu za standardima. JavaScript je sličan. JavaScript je dijalekt, a ne standard. Standard je ECMAScript , stvorio ECMA - Europsko udruženje proizvođača računala. ECMAScript je pokušaj standardizacije JavaScript-a.

Postoji više od jedne implementacije ECMAScripta, ali JavaScript je najpopularniji, pa se stoga naziv JavaScript i ECMAScript često koriste naizmjenično.

JavaScript se pokreće u stroju.

JavaScript je samo tekstualna datoteka. Poput vozača bez automobila, ne može se ići daleko. Nešto mora pokrenuti ili protumačiti vašu datoteku. To radi JavaScript mehanizam.

Nekoliko primjera JavaScript motora uključuju V8, mehanizam koji koristi Google Chrome; SpiderMonkey, motor koji koristi Mozilla Firefox; i JavaScriptCore, motor koji koristi Apple Safari. ECMAScript, jezični standard, osigurava dosljednost različitih JavaScript mehanizama.

JavaScript motori rade u okruženju.

Iako se JavaScript može izvoditi na različitim mjestima (na primjer, Node.js, popularna tehnologija na strani poslužitelja, pokreće JavaScript i koristi isti V8 mehanizam koji koristi Google Chrome), najčešće mjesto za pronalaženje JavaScript mehanizma je web preglednik.

Unutar preglednika, JavaScript mehanizam samo je jedan dio većeg okruženja koji pomaže oživjeti naš kôd. Tri su glavna dijela ovog okruženja i zajedno čine ono što se naziva runtime okruženje .

Niz poziva

Prvi dio je mjesto trenutno pokrenutog koda. To se naziva skupom poziva. U JavaScriptu postoji samo jedan skup poziva, a to će postati važno kako nastavljamo graditi svoje temelje.

Evo pojednostavljenog primjera niza poziva:

function doSomething() { //some other code doSomethingElse(); //some other code } function doSomethingElse() { //some other code } doSomething();

Inicijalni niz poziva je prazan, jer ne postoji pokrenut kôd. Kada naš JavaScript motor konačno dosegne prvi poziv funkcije doSomething(), on se dodaje u stog:

--Call Stack-- doSomething;

Unutar programa doSomething()pokrećemo neki drugi kôd i zatim dolazimodoSomethingElse():

--Call Stack-- doSomething doSomethingElse

Kada doSomethingElse()se završi, uklanja se iz niza poziva:

--Call Stack-- doSomething

Konačno, doSomething()završava preostali kôd i uklanja se iz niza poziva:

--Call Stack-- Empty

Web API-ji

Drugi dio našeg okruženja preglednika popunjava donekle prazninu. Iznenađujuće, stvari poput interakcije s DOM-om, izrade zahtjeva za poslužiteljem i većine zadataka temeljenih na pregledniku nisu dio jezičnog standarda ECMAScript.

Srećom, preglednici nam nude dodane značajke na koje se naš JavaScript motor može uključiti. Te značajke proširuju funkcionalnost JavaScript-a unutar preglednika. Omogućuju nam radnje poput preslušavanja događaja ili izrade zahtjeva poslužitelja - stvari koje JavaScript ne može sam. I zovu se w eb API-ji .

Mnogi web API-ji omogućuju nam preslušavanje ili čekanje da se nešto dogodi. Kada se dogodi taj događaj, tada pokrećemo neki drugi kod.

Evo našeg primjera stoga poziva proširenog kako bi obuhvatio (pretvaranje) web API.

function doSomething() { //some other code listenForClick(); doSomethingElse(); //some other code } function doSomethingElse() { //some other code } listenForClick() { console.log('the button was clicked!') } doSomething();

Kada se preglednik susretne doSomething(), smjesti se u niz poziva:

--Call Stack-- doSomething

Zatim pokreće neki drugi kôd i zatim nailazi listenForClick(...):

--Call Stack-- doSomething listenForClick

listenForClick() se priključuje na web API i u ovom slučaju uklanja se iz niza poziva.

JavaScript motor sada prelazi na doSomethingElse():

--Call Stack-- doSomething doSomethingElse

doSomethingElse()i doSomething()završi, a skup poziva je prazan. Ali što se dogodilo listenForClick()?

Red čekanja za događaje

This is where we introduce the final part of our browser environment. Often, our web API code is a function that takes a callback. A callback is just some code we want to run after another function runs. For example, listening for a click event and then console.log something. In order to make sure our console.log doesn’t interfere with any currently running code, it first passes to something called an event queue.

The event queue acts as a waiting area until our call stack is empty. Once the call stack is empty, the event queue can pass our code into the call stack to be run. Let’s continue to build upon our previous example:

function doSomething() { //some other code listenForClick(); doSomethingElse(); //some other code } function doSomethingElse() { //some other code } listenForClick() { console.log('the button was clicked!') } doSomething();

So now, our code runs like this:

Our engine encounters doSomething():

--Call Stack-- doSomething

doSomething() runs some code and then encounters listenForClick(...). In our example, this takes a callback, which is the code we want to run after the user clicks a button. The engine passes listenForClick(…) out of the call stack and continues until it encounters doSomethingElse():

--Call Stack-- doSomething doSomethingElse

doSomethingElse() runs some code, and finishes. By this time, our user clicks the button. The web API hears the click and sends the console.log() statement to the event queue. We’ll pretend doSomething() is not done; therefore, the call stack is not empty, and the console.log() statement must wait in the event queue.

--Call Stack-- doSomething

After a few seconds, doSomething() finishes and is removed from the call stack:

--Call Stack-- EMPTY

Finally, the console.log() statement can get passed into the call stack to be executed:

--Call Stack-- console.log('The user clicked the button!')

Keep in mind, our code is running incredibly fast — taking single-digit milliseconds to finish. It isn’t realistic we could start our code, and our user could click a button before the code is done running. But in our simplified example, we pretend that this is true, to highlight certain concepts.

Together, all three parts (the call stack, the web APIs, and the event queue) form what is called the concurrency model, with the event loop managing the code that goes from the event queue into the call stack.

Take aways from the above examples:

JavaScript can only do one thing at a time.

There is a misconception that people can multi-task. This isn’t true. People can, however, switch between tasks, a process called task switching.

JavaScript is similar in the sense that it can’t multitask. Because JavaScript has only one call stack, the JavaScript engine can only do one task at a time. We say this makes JavaScript single threaded. Unlike people, however, JavaScript can’t task switch without the help of our web APIs.

JavaScript must finish a task before moving on.

Because JavaScript can’t switch back and forth between tasks, if you have any code that takes a while to run, it will block the next line of code from running. This is called blocking code, and it happens because JavaScript is synchronous. Synchronous simply means that JavaScript must finish a task before it can start another one.

An example of blocking code might be a server request which requires us to wait for data to be returned. Fortunately, the web APIs provided by the browser give us a way around this (with the use of callbacks).

By moving blocking code from the call stack into the event loop, our engine can move on to the next item in the call stack. Therefore, with code running in our call stack, and code that is simultaneously running in a web API, we have asynchronous behavior.

Not all web APIs, however, go into the event loop. For example, console.log is a web API, but since it has no callback and doesn’t need to wait for anything, it can be executed immediately.

Do keep in mind that single threaded is not the same as synchronous. Single threaded means “one thing at a time.” Synchronous means “finish before moving on.” Without the help of asynchronous APIs, core JavaScript is both single threaded and synchronous.

The scoop on scope

Before we return to our original question, we need to touch on scope. Scope is the term used to describe which parts of our code have access to which variables.

Intuitively, it may seem that a variable declared and initialized by a for loop would only be available within that for loop. In other words, if you tried to access it outside of the loop, you would get an error.

This isn’t the case. Declaring a variable with the varkeyword creates a variable that is also available in its parent scope.

This example shows that a variable declared by var within a for loop is also available within the parent scope (in this case, the global scope).

for(var a = 1; a < 10; a++) {} // declared "inside" the loop console.log(a); // prints "10" and is called "outside the loop"

The answer revealed

At this point, we’ve discussed enough to build our answer.

Here is our example revisited:

for(var i = 1; i < 6; i++) { setTimeout(function() { console.log(i); },1000); } console.log('The loop is done!');

Intuitively, you might believe this will print the numbers one through five, with one second between each number being printed:

// one second between each log 1 2 3 4 5 The loop is done!

However, what we actually output is:

The loop is done! // then about one second later and all at once 6 6 6 6 6

What’s happening?

Recall our discussion about web APIs. Asynchronous web API’s, or those with callbacks, go through the event loop. setTimeout()happens to be an asynchronous web API.

Every time we loop, setTimeout() is passed outside of the call stack and enters the event loop. Because of this, the engine is able to move to the next piece of code. The next piece of code happens to be the remaining iterations of the loop, followed by console.log(‘The loop is done!’).

To show the setTimeout() statements are being passed from the call stack, and the loop is running, we can place a console.log() statement outside of the setTimeout() function and print the results. We can also place a built-in timer method to show just how quickly everything is happening. We use console.time() and console.timeEnd() to do this .

console.time('myTimer'); for(var i = 1; i { console.log(i); },1000); } console.log('The loop is done!'); console.timeEnd('myTimer');

Results:

Loop Number 1 Loop Number 2 Loop Number 3 Loop Number 4 Loop Number 5 The loop is done! // then, about one second later and all at once: 6 6 6 6 6 myTimer: 1.91577ms // Wow, that is quick!

First, we can see the loop is in fact running. In addition, the timer we added tells us that everything other than our setTimeout() functions took less than two milliseconds to run! That means each setTimeout() function has about 998 milliseconds remaining before the code it contains goes into the event queue and then finally into the call stack. Remember earlier when I said it would be difficult for a user to be faster than our code!

If you run this code multiple times, you will likely notice the timer output will change slightly. This is because your computer’s available resources are always changing and it might be slightly faster or slower each time.

So here is what’s happening:

  1. Our engine comes across our for loop. We declare and initialize a global variable named i equal to one.
  2. Each iteration of loop passes setTimeout() to a web API and into the event loop. Therefore, our for loop finishes very quickly, since there is no other code inside of it to run. In fact, the only thing our loop does is change the value of i to six.
  3. At this point, the loop is over, our setTimeout() functions are still counting down, and all that remains in the call stack is console.log(‘The loop is done!’).
  4. Fast forward a bit, and the setTimeout() functions have finished, and the console.log(i) statements go into the event queue. By this time, our console.log(‘The loop is done!’) has been printed and the call stack is empty.
  5. Since the call stack is empty, the fiveconsole.log(i) statements get passed from the event queue into the call stack.
  6. Remember, i is now equal to six, and that’s why we see five sixes printed to the screen.

Let’s create the output we thought we would get

Up to this point, we’ve discussed the actual output of a few simple lines of code that turned out to be not-so-simple. We’ve talked about what’s happening on a deeper level and what the result is. But, what if we want to create the output we thought we would get? In other words, how can we reverse engineer the following results:

1 // after one second, then 2 // one second later (2 seconds total) 3 // one second later (3 seconds total) 4 // one second later (4 seconds total) 5 // one second later (5 seconds total) 'The loop is done!' // one second later (6 seconds total)

Does the duration on our timeout change anything?

Setting the timeout’s duration to zero seems like a possible solution. Let’s give it a try.

for(var i = 1; i { console.log(i); },0); } console.log('The loop is done!');

Results:

// Everything appears (essentially) at once The loop is done! 6 6 6 6 6

It still didn’t work. What happened?

Remember, just because the duration of setTimeout() is zero, it is still asynchronous and handled by a web API. Regardless of the duration, it will be passed to the event queue and then the call stack. So even with a timeout of zero, the process remains the same, and the output is relatively unchanged.

Notice I said relatively. One thing you may have noticed that was different, was everything printed almost at once. This is because the duration of setTimeout() expires instantly, and its code gets from the web API, into the event queue, and finally into the call stack almost immediately. In our previous example, our code had to wait 1000 milliseconds before it went into the event queue and then the call stack.

So, if changing the duration to zero didn’t work, now what?

Revisiting Scope

What will this code output?

 function myFunction1() { var a = 'Brandon'; console.log(a); } function myFunction2() { var a = 'Matt'; console.log(a); } function myFunction3() { var a = 'Bill'; console.log(a); } myFunction1() myFunction2() myFunction3()

Notice how each function uses the same variable named a. It would seem each function might throw an error, or possibly overwrite the value of a.

Results:

Brandon Bill Matt

There is no error, and a is unique each time.

It appears the variable a is unique to each function. It’s very similar to how an address works. Street names and numbers are invariably shared across the world. There is more than a single 123 Main St. It’s the city and state which provide scope to which address belongs where.

Functions work in the same way. Functions act as a protective bubble. Anything inside of that bubble can’t be accessed by anything outside. This is why the variable a is not actually the same variable. It’s three different variables located in three different places in memory. They just so happen to all share the same name.

Applying the principles of scope to our example:

We know we have access to the iterative value of i, just not when the setTimeout() statements finish. What if we take the value of i and package it with the setTimeout() statement in its own bubble (as a way to preserve i)?

for(var i = 1; i <6; i++) { function timer(){ // create a unique function (scope) each time var k = i; // save i to the variable k which setTimeout(()=>{ console.log(k); },1000); } timer(); }

Result:

The loop is done! 1 2 3 4 5

It almost works. What did we do?

We are beginning to get into the topic of closures. A deep discussion on closures goes beyond the scope of this article. However, a brief introduction will help our understanding.

Remember, each function creates a unique scope. Because of this, variables with the same name can exist in separate functions and not interfere with each other. In our most recent example, each iteration created a new and unique scope (along with a new and unique variable k). When the for loop is done, these five unique values of k are still in memory and are accessed appropriately by our console.log(k) statements. That is closure in a nutshell.

In our original example where we declare i with var, each iteration overwrote the value of i (which in our case was a global variable).

ES6 makes this much cleaner.

In 2015, ECMAScript released a major update to its standards. The update contained many new features. One of those features was a new way to declare variables. Up to this point, we have used the var keyword to declare variables. ES6 introduced the let keyword.

for(let i = 1; i { console.log(i); },1000); } console.log('The loop is done!');

Results:

The loop is done! 1 2 3 4 5

Just by changing var to let, we are much closer to the result we want.

A brief introduction to “let” vs “var”

In our example, let does two things:

First, it makes i available only inside our for loop. If we try to log i outside of the loop, we get an error. This is because let is a block scope variable. If it is inside a block of code (such as a for loop) it can only be accessed there. var is function scoped.

An example to show let vs var behavior:

function variableDemo() { var i = 'Hello World!'; for(let i = 1; i < 3; i++) { console.log(i); // 1, 2, 3 } console.log(i); // "Hello World!" // the for-loop value of i is hidden outside of the loop with let } variableDemo(); console.log(i); //Error, can't access either value of i

Notice how we don’t have access to either i outside of the function variableDemo(). This is because ‘Hello World’ is function scoped, and i is block scoped.

The second thing let does for us is create a unique value of i each time the loop iterates. When our loop is over, we have created six separate values of i that are stored in memory that our console.log(i) statements can access. With var, we only had one variable that we kept overwriting.

The loop is not done.

We’re almost there. We still are logging 'The loop is done!' first, and we aren’t logging everything one second apart. First, we will look at two ways to address the The loop is done! output.

Option 1: Using setTimeout() and the concurrency model to our advantage.

This is fairly straightforward. We want The loop is done! to pass through the same process as the console.log(i) statements. If we wrap The loop is done! in a setTimeout() whose duration is greater to or equal than the for loop timeouts, we ensure The loop is done! arrives behind and expires after the last for loop timeouts.

We’ll break up our code a bit to make it a bit clearer:

function loopDone() { // we will call this below console.log('The loop is done!)' } for(let i = 1; i { console.log(i); },1000); } setTimeout(loopDone, 1001);

Results:

1 2 3 4 5 The loop is done!

Option 2: Check for the final console.log(i) completion

Another option is to check when the console.log(i) statements are done.

function loopDone() { console.log('The loop is done!'); } for(let i = 1; i { console.log(i); if(i === 5){ // check when the last statement has been logged loopDone(); } },1000); }

Results:

1 2 3 4 5 The loop is done!

Notice that we placed our loop completion check within the setTimeout() function, not within the main body of the for loop.

Checking when the loop is done won’t help us, since we still must wait for the timeouts to complete. What we want to do is check when the console.log(i) statements are done. We know this will be after the value of i is 5 and after we’ve logged it. If we place our loop completion check after the console.log(i) statement, we can ensure we’ve logged the final ibefore we run loopDone().

Getting everything to happen one second apart.

Everything is happening essentially at the same time because the loop is so fast, and all the timeouts arrive at the web API within milliseconds of each other. Therefore, they expire around the same time and go to the event queue and call stack around the same time.

We can’t easily change when they arrive at the web API. But we can, with the unique value of each i, delay how long they stay there.

function loopDone() { console.log('The loop is done!'); } for(let i = 1; i { console.log(i); if(i === 5){ loopDone(); } },i * 1000); // multiple i by 1000 }

Since i is now unique (because we are using let), if we multiply i by 1000, each timeout will last one second longer than the previous timeout. The first timeout will arrive with a 1000 millisecond duration, the second with 2000 and so forth.

Although they arrive at the same time, it will now take each timeout one second longer than the previous to pass to the event queue. Since our call stack is empty by this point, it goes from the event queue immediately into the call stack to be executed. With each console.log(i) statement arriving one second apart in the event queue, we will almost have our desired output.

1 // after one second, then 2 // one second later (2 seconds total) 3 // one second later (3 seconds total) 4 // one second later (4 seconds total) 5 // one second later (5 seconds total) 'The loop is done!' // still occurs with the final log

Notice that The loop is done! is still arriving with the last console.log(i) statement, not one second after it. This is because when i===5loopDone() is run. This prints both the i and The loop is done! statements around the same time.

We can simply wrap loopDone() in a setTimeout() to address this.

function loopDone() { console.log('The loop is done!'); } for(let i = 1; i { console.log(i); if(i === 5){ setTimeout(loopDone, 1000); // update this } },i * 1000); }

Results:

1 // after one second, then 2 // one second later (2 seconds total) 3 // one second later (3 seconds total) 4 // one second later (4 seconds total) 5 // one second later (5 seconds total) 'The loop is done!' // one second later (6 seconds total)

We finally have the results we wanted!

Most of this article stemmed from my own struggles and the subsequent aha! moments in an attempt to understand closures and the JavaScript event loop. I hope this can make sense of the basic processes at play and serve as a foundation for more advanced discussions of the topic.

Thanks!

woz