Is onload equal to readyState==4 in XMLHttpRequest?

I am confuse about the xhr return event, as I can tell, there are not so much different between onreadystatechange --> readyState == 4 and onload, is it true?

var xhr = new XMLHttpRequest();
xhr.open("Get", url, false);
xhr.onreadystatechange = function() {
    if (xhr.readyState === 4)
    {
        /* do some thing*/
    }
};

xhr.send(null);

or

xhr.onload = function() { /* do something */ }

This is almost always true. One significant difference, however, is that the onreadystatechange event handler also gets triggered with readyState==4 in the cases where the onerror handler is usually triggered (typically a network connectivity issue). It gets a status of 0 in this case. I've verified this happens on the latest Chrome, Firefox and IE.

So if you are using onerror and are targeting modern browsers, you should not use onreadystatechange but should use onload instead, which seems to be guaranteed to only be called when the HTTP request has successfully completed (with a real response and status code). Otherwise you may end up getting two event handlers triggered in case of errors (which is how I empirically found out about this special case.)

Here is a link to a Plunker test program I wrote that lets you test different URLs and see the actual sequence of events and readyState values as seen by the JavaScript app in different cases. The JS code is also listed below:

var xhr;
function test(url) {
    xhr = new XMLHttpRequest();
    xhr.addEventListener("readystatechange", function() { log(xhr, "readystatechange") });
    xhr.addEventListener("loadstart", function(ev) { log(xhr, "loadstart", ev.loaded + " of " + ev.total) });
    xhr.addEventListener("progress", function(ev) { log(xhr, "progress", ev.loaded + " of " + ev.total) });
    xhr.addEventListener("abort", function() { log(xhr, "abort") });
    xhr.addEventListener("error", function() { log(xhr, "error") });
    xhr.addEventListener("load", function() { log(xhr, "load") });
    xhr.addEventListener("timeout", function(ev) { log(xhr, "timeout", ev.loaded + " of " + ev.total) });
    xhr.addEventListener("loadend", function(ev) { log(xhr, "loadend", ev.loaded + " of " + ev.total) });
    xhr.open("GET", url);
    xhr.send();
}

function clearLog() {
    document.getElementById('log').innerHTML = '';
}

function logText(msg) {
    document.getElementById('log').innerHTML += msg + "<br/>";
}

function log(xhr, evType, info) {
    var evInfo = evType;
    if (info)
        evInfo += " - " + info ;
    evInfo += " - readyState: " + xhr.readyState + ", status: " + xhr.status;
    logText(evInfo);
}

function selected(radio) {
    document.getElementById('url').value = radio.value;
}

function testUrl() {
    clearLog();
    var url = document.getElementById('url').value;
    if (!url)
        logText("Please select or type a URL");
    else {
        logText("++ Testing URL: " + url);
        test(url);
    }
}

function abort() {
    xhr.abort();
}

It should be the same thing. onload was added in XMLHttpRequest 2 whereas onreadystatechange has been around since the original spec.

No, they are not the same. If you encounter a network error or abort the operation, onload will not be called. Actually, the closest event to readyState === 4 would be loadend. The flow looks like this:

     onreadystatechange
      readyState === 4
             ⇓
 onload / onerror / onabort
             ⇓
         onloadend

in simple code here how they are handle the error

xhr.onload = function() {
  // same or allowed cross origin
  if (this.status == 200) {

  }
  else {} // error http status not 200
};
xhr.onerror = function() {
  //error: cross origin, bad connection
};

VS

xhr.onreadystatechange = function() {
  if (xhr.readyState === 4) {
    if (this.status == 200) {

    }
    else {} // error: cross origin, http status not 200, bad connection
  }
};