Develop Reference

How does inline Javascript (in HTML) work?

I know this is bad practice. Don't write code like this if at all possible.
Of course, we'll always find ourselves in situations where a clever snippet of inline Javascript can address an issue quickly.
I am pursuing this query in the interest of fully understanding what happens (and the potential pitfalls) when something like this is written:
Click Me
As far as I can tell this is functionally the same as
<script type="text/javascript">
$(function(){ // I use jQuery in this example
document.getElementById('click_me').onclick =
function () { alert('Hi'); };
});
</script>
Click Me
Extrapolating from this it seems that the string assigned to attribute onclick is inserted within an anonymous function which is assigned to the element's click handler. Is this actually the case?
Because I'm starting to do things like this:
Display my next sibling <!-- Return false in handler so as not to scroll to top of page! -->
Which works. But I don't know how much of a hack this is. It looks suspicious because there is no apparent function that is being returned from!
You might ask, why are you doing this, Steve? Inline JS is bad practice!
Well to be quite honest I'm tired of editing three different sections of code just to modify one section of a page, especially when I'm just prototyping something to see if it will work at all. It is so much easier and sometimes even makes sense for the code specifically related to this HTML element to be defined right within the element: When I decide 2 minutes later that this was a terrible, terrible idea I can nuke the entire div (or whatever) and I don't have a bunch of mysterious JS and CSS cruft hanging around in the rest of the page, slowing down rendering ever so slightly. This is similar to the concept of locality of reference but instead of cache misses we're looking at bugs and code bloat.

You've got it nearly correct, but you haven't accounted for the this value supplied to the inline code.
Click Me
is actually closer to:
Click Me
<script type="text/javascript">
document.getElementById('click_me').addEventListener("click", function(event) {
(function(event) {
alert(this);
}).call(document.getElementById('click_me'), event);
});
</script>
Inline event handlers set this equal to the target of the event.
You can also use anonymous function in inline script
Click Me

What the browser does when you've got
<a onclick="alert('Hi');" ... >
is to set the actual value of "onclick" to something effectively like:
new Function("event", "alert('Hi');");
That is, it creates a function that expects an "event" parameter. (Well, IE doesn't; it's more like a plain simple anonymous function.)

There seems to be a lot of bad practice being thrown around Event Handler Attributes. Bad practice is not knowing and using available features where it is most appropriate. The Event Attributes are fully W3C Documented standards and there is nothing bad practice about them. It's no different than placing inline styles, which is also W3C Documented and can be useful in times. Whether you place it wrapped in script tags or not, it's gonna be interpreted the same way.
https://www.w3.org/TR/html5/webappapis.html#event-handler-idl-attributes

The best way to answer your question is to see it in action.
<a id="test" onclick="alert('test')"> test </a> ​
In the js
var test = document.getElementById('test');
console.log( test.onclick );
As you see in the console, if you're using chrome it prints an anonymous function with the event object passed in, although it's a little different in IE.
function onclick(event) {
alert('test')
}
I agree with some of your points about inline event handlers. Yes they are easy to write, but i don't agree with your point about having to change code in multiple places, if you structure your code well, you shouldn't need to do this.

It looks suspicious because there is no apparent function that is being returned from!
It is an anonymous function that has been attached to the click event of the object.
why are you doing this, Steve?
Why on earth are you doi.....Ah nevermind, as you've mentioned, it really is widely adopted bad practice :)

Try this in the console:
var div = document.createElement('div');
div.setAttribute('onclick', 'alert(event)');
div.onclick
In Chrome, it shows this:
function onclick(event) {
alert(event)
}
...and the non-standard name property of div.onclick is "onclick".
So, whether or not this is anonymous depends your definition of "anonymous." Compare with something like var foo = new Function(), where foo.name is an empty string, and foo.toString() will produce something like
function anonymous() {
}

using javascript:
here input element is used
<input type="text" id="fname" onkeyup="javascript:console.log(window.event.key)">
if you want to use multiline code use curly braces after javascript:
<input type="text" id="fname" onkeyup="javascript:{ console.log(window.event.key); alert('hello'); }">

Change background "onfocus" + other things

I have the script
<input type="text" name="name" value="Joe Bloggs" onfocus="this.value=''"/>
Just like I am doing with the onfocus="this.value…" can I change the background of the field (+ change other things?)
Also, does anyone have a better idea of how to do the script above, but so that the text reappears when the deselect it?

It's considered best practice to leave styling to CSS and logic to JS.
You can do this in css using the :focus pseudo-class. (http://www.quirksmode.org/css/focus.html) Unfortunately, it doesn't work in IE7 or lower. For those browsers, you can use javascript to add a class to the <input> to do the same thing.
CSS
input:focus, input.focus {background:#ff0}
If you are using jQuery, here is how you would do that.
$('input').focus(function(){
$(this).addClass('focus');
});
$('input').blur(function(){
$(this).removeClass('focus');
});

Something along the lines of
onfocus="this.value=''; this.style.backgroundColor='#f00'" onblur="this.style.backgroundColor='white'"
will get approximately what you want done simply, although it would also be possible for it to interact badly with presentation defined elsewhere and as such is probably rather crude to be considered a best practise.
Alternatively, you could as suggested add / remove a specific class to the element onfocus / onblur. At this point I would also second the jQuery recommendation: although it's hardly necessary just for this, you will find that it makes life with Javascript in general much more pleasant.
If you use jQuery, something like
$('input').focus(function() { $(this).addClass('focus') });
$('input').blur(function() { $(this).removeClass('focus') });
would allow you to cleanly define the appearance of focussed inputs in CSS. Consult jQuery documentation for the surrounding context necessary to make this work.

...id="name" onfocus = "javascript:yourFunction();"...
Then, your js looks something like this:
var inputFld = document.getElementById('name');<br/>
var oldval = inputFld.value;<br/>
var tempval = "";<br/>
function yourFunction(){
inputFld.value = tempval;<br/>
inputFld.className = "test" //(building from the first answer)<br/>
//do other stuff...<br/>
}
Then, you can also add an onblur="javascript:anotherFunction();" where anotherFunction() resets inputFld.value to the original value.
Note, best practices advise you should avoid those global vars and that you attach event listeners for the onblur and onfocus events rather than inlining them. But, at least, initially, you can see if this code as it is written works for you...

On your event, append the class name...
el.className+= 'test'
then set a background on that class in your css.

Can I declare the same value for two attributes at once

For example, for accessibility reasons I want my onfocus and onmouseover to have identical values. For the sake of maintainability I want to declare this once only.
What I'd like to be able to do is this:
<a onmouseover,onfocus="assist('hello');">linktext</a>
But, of course, that would be too easy, and doesn't work. What's the best way I can achieve this DRYing out of my tags?

The better way is to move all your event handlers out of the HTML and into javascript.
<a id="some_id">link text</a>
(You don't need to use an ID, it's just for ease of demonstration.)
var link = document.getElementById('some_id');
link.onmouseover = link.onfocus = function() {
assist('hello');
};
If you wanted to take it to the next level, you could generalise it across many links. This example uses jQuery.
link text
another link
Then you apply the event handler to all the links in one go (very dry!)
$('a.assist').bind('mouseover focus', function() {
assist(this.href); // you'd need to clean up the href here, but you get the picture
});

Best I can think of is to use a function, set both events to call the function and put the common code in there.

determine when contents of html tag changes

Is there a way to determine when the contents of a HTML tag has changed? I would prefer to catch an event rather than polling it.
My use case is I have text enclosed in span tags within a rich text editor, and I need to remove the span tags when the enclosing text is modified by the user.

Are you using one of the typical WYSIWYG editors, and don't want to update their code to break updatability? Then maybe you could listen to the onTextChange event (or something similar) that the WYSIWYG editor is sending, check the contents of the change, and react on that.
Just an idea, given that you give a bit too little information in your question.

I don't think there is an event available (except for input elements), however you could poll it.
$element = $('#my-element');
var originalHtml = $element.html();
var pollHtml = setInterval(function() {
if (originalHtml !== $element.html()) {
alert('HTML changed!');
clearInterval(pollHtml);
};
}, 100);

I can't add a comment so I'm putting my suggestion here,
#Justin Johnson
$("#close-question-2114317").change(function() {alert(1);});
$("#close-question-2114317").text( function(){
$(this).trigger('change');
return "Close!!!";
});
I believe we can call the trigger if that's the case...

You must be using JavaScript to modify the DOM in some way. Polling is in most cases a bad idea. I have set up a demo here that gives you a good idea on how you might want to check the HTML for changes. Just call that function when needed... even if it is just polling.

How to store arbitrary data for some HTML tags

I'm making a page which has some interaction provided by javascript. Just as an example: links which send an AJAX request to get the content of articles and then display that data in a div. Obviously in this example, I need each link to store an extra bit of information: the id of the article. The way I've been handling it in case was to put that information in the href link this:
<a class="article" href="#5">
I then use jQuery to find the a.article elements and attach the appropriate event handler. (don't get too hung up on the usability or semantics here, it's just an example)
Anyway, this method works, but it smells a bit, and isn't extensible at all (what happens if the click function has more than one parameter? what if some of those parameters are optional?)
The immediately obvious answer was to use attributes on the element. I mean, that's what they're for, right? (Kind of).
<a articleid="5" href="link/for/non-js-users.html">
In my recent question I asked if this method was valid, and it turns out that short of defining my own DTD (I don't), then no, it's not valid or reliable. A common response was to put the data into the class attribute (though that might have been because of my poorly-chosen example), but to me, this smells even more. Yes it's technically valid, but it's not a great solution.
Another method I'd used in the past was to actually generate some JS and insert it into the page in a <script> tag, creating a struct which would associate with the object.
var myData = {
link0 : {
articleId : 5,
target : '#showMessage'
// etc...
},
link1 : {
articleId : 13
}
};
<a href="..." id="link0">
But this can be a real pain in butt to maintain and is generally just very messy.
So, to get to the question, how do you store arbitrary pieces of information for HTML tags?

Which version of HTML are you using?
In HTML 5, it is totally valid to have custom attributes prefixed with data-, e.g.
<div data-internalid="1337"></div>
In XHTML, this is not really valid. If you are in XHTML 1.1 mode, the browser will probably complain about it, but in 1.0 mode, most browsers will just silently ignore it.
If I were you, I would follow the script based approach. You could make it automatically generated on server side so that it's not a pain in the back to maintain.

If you are using jQuery already then you should leverage the "data" method which is the recommended method for storing arbitrary data on a dom element with jQuery.
To store something:
$('#myElId').data('nameYourData', { foo: 'bar' });
To retrieve data:
var myData = $('#myElId').data('nameYourData');
That is all that there is to it but take a look at the jQuery documentation for more info/examples.

Just another way, I personally wouldn't use this but it works (assure your JSON is valid because eval() is dangerous).
<a class="article" href="link/for/non-js-users.html">
<span style="display: none;">{"id": 1, "title":"Something"}</span>
Text of Link
</a>
// javascript
var article = document.getElementsByClassName("article")[0];
var data = eval(article.childNodes[0].innerHTML);

Arbitrary attributes are not valid, but are perfectly reliable in modern browsers. If you are setting the properties via javascript, than you don't have to worry about validation as well.
An alternative is to set attributes in javascript. jQuery has a nice utility method just for that purpose, or you can roll your own.

A hack that's going to work with pretty much every possible browser is to use open classes like this: <a class='data\_articleid\_5' href="link/for/non-js-users.html>;
This is not all that elegant to the purists, but it's universally supported, standard-compliant, and very easy to manipulate. It really seems like the best possible method. If you serialize, modify, copy your tags, or do pretty much anything else, data will stay attached, copied etc.
The only problem is that you cannot store non-serializable objects that way, and there might be limits if you put something really huge there.
A second way is to use fake attributes like: <a articleid='5' href="link/for/non-js-users.html">
This is more elegant, but breaks standard, and I'm not 100% sure about support. Many browsers support it fully, I think IE6 supports JS access for it but not CSS selectors (which doesn't really matter here), maybe some browsers will be completely confused, you need to check it.
Doing funny things like serializing and deserializing would be even more dangerous.
Using ids to pure JS hash mostly works, except when you try to copy your tags. If you have tag <a href="..." id="link0">, copy it via standard JS methods, and then try to modify data attached to just one copy, the other copy will be modified.
It's not a problem if you don't copy tags, or use read only data. If you copy tags and they're modified you'll need to handle that manually.

Using jquery,
to store: $('#element_id').data('extra_tag', 'extra_info');
to retrieve: $('#element_id').data('extra_tag');

I know that you're currently using jQuery, but what if you defined the onclick handler inline. Then you could do:
<a href='/link/for/non-js-users.htm' onclick='loadContent(5);return false;'>
Article 5</a>

You could use hidden input tags. I get no validation errors at w3.org with this:
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html lang='en' xml:lang='en' xmlns='http://www.w3.org/1999/xhtml'>
<head>
<meta content="text/html;charset=UTF-8" http-equiv="content-type" />
<title>Hello</title>
</head>
<body>
<div>
<a class="article" href="link/for/non-js-users.html">
<input style="display: none" name="articleid" type="hidden" value="5" />
</a>
</div>
</body>
</html>
With jQuery you'd get the article ID with something like (not tested):
$('.article input[name=articleid]').val();
But I'd recommend HTML5 if that is an option.

Why not make use of the meaningful data already there, instead of adding arbitrary data?
i.e. use <a href="/articles/5/page-title" class="article-link">, and then you can programmatically get all article links on the page (via the classname) and the article ID (matching the regex /articles\/(\d+)/ against this.href).

As a jQuery user I would use the Metadata plugin. The HTML looks clean, it validates, and you can embed anything that can be described using JSON notation.

This is good advice. Thanks to #Prestaul
If you are using jQuery already then you should leverage the "data"
method which is the recommended method for storing arbitrary data on a
dom element with jQuery.
Very true, but what if you want to store arbitrary data in plain-old HTML? Here's yet another alternative...
<input type="hidden" name="whatever" value="foobar"/>
Put your data in the name and value attributes of a hidden input element. This might be useful if the server is generating HTML (i.e. a PHP script or whatever), and your JavaScript code is going to use this information later.
Admittedly, not the cleanest, but it's an alternative. It's compatible with all
browsers and is valid XHTML. You should NOT use custom attributes, nor should you really use attributes with the 'data-' prefix, as it might not work on all browsers. And, in addition, your document will not pass W3C validation.

As long as you're actual work is done serverside, why would you need custom information in the html tags in the output anyway? all you need to know back on the server is an index into whatever kind of list of structures with your custom info. I think you're looking to store the information in the wrong place.
I will recognize, however unfortunate, that in lots of cases the right solution isn't the right solution. In which case I would strongly suggest generating some javascript to hold the extra information.
Many years later:
This question was posted roughly three years before data-... attributes became a valid option with the advent of html 5 so the truth has shifted and the original answer I gave is no longer relevant. Now I'd suggest to use data attributes instead.
<a data-articleId="5" href="link/for/non-js-users.html">
<script>
let anchors = document.getElementsByTagName('a');
for (let anchor of anchors) {
let articleId = anchor.dataset.articleId;
}
</script>

I advocate use of the "rel" attribute. The XHTML validates, the attribute itself is rarely used, and the data is efficiently retrieved.

So there should be four choices to do so:
Put the data in the id attribute.
Put the data in the arbitrary attribute
Put the data in class attribute
Put your data in another tag
http://www.shanison.com/?p=321

You could use the data- prefix of your own made attribute of a random element (<span data-randomname="Data goes here..."></span>), but this is only valid in HTML5. Thus browsers may complain about validity.
You could also use a <span style="display: none;">Data goes here...</span> tag. But this way you can not use the attribute functions, and if css and js is turned off, this is not really a neat solution either.
But what I personally prefer is the following:
<input type="hidden" title="Your key..." value="Your value..." />
The input will in all cases be hidden, the attributes are completely valid, and it will not get sent if it is within a <form> tag, since it has not got any name, right?
Above all, the attributes are really easy to remember and the code looks nice and easy to understand. You could even put an ID-attribute in it, so you can easily access it with JavaScript as well, and access the key-value pair with input.title; input.value.

One possibility might be:
Create a new div to hold all the extended/arbitrary data
Do something to ensure that this div is invisible (e.g. CSS plus a class attribute of the div)
Put the extended/arbitrary data within [X]HTML tags (e.g. as text within cells of a table, or anything else you might like) within this invisible div

Another approach can be to store a key:value pair as a simple class using the following syntax :
<div id="my_div" class="foo:'bar'">...</div>
This is valid and can easily be retrieved with jQuery selectors or a custom made function.

In html, we can store custom attributes with the prefix 'data-' before the attribute name like
<p data-animal='dog'>This animal is a dog.</p>.
Check documentation
We can use this property to dynamically set and get attributes using jQuery like:
If we have a p tag like
<p id='animal'>This animal is a dog.</p>
Then to create an attribute called 'breed' for the above tag, we can write:
$('#animal').attr('data-breed', 'pug');
To retrieve the data anytime, we can write:
var breedtype = $('#animal').data('breed');

At my previous employer, we used custom HTML tags all the time to hold info about the form elements. The catch: We knew that the user was forced to use IE.
It didn't work well for FireFox at the time. I don't know if FireFox has changed this or not, but be aware that adding your own attributes to HTML elements may or may-not be supported by your reader's browser.
If you can control which browser your reader is using (i.e. an internal web applet for a corporation), then by all means, try it. What can it hurt, right?

This is how I do you ajax pages... its a pretty easy method...
function ajax_urls() {
var objApps= ['ads','user'];
$("a.ajx").each(function(){
var url = $(this).attr('href');
for ( var i=0;i< objApps.length;i++ ) {
if (url.indexOf("/"+objApps[i]+"/")>-1) {
$(this).attr("href",url.replace("/"+objApps[i]+"/","/"+objApps[i]+"/#p="));
}
}
});
}
How this works is it basically looks at all URLs that have the class 'ajx' and it replaces a keyword and adds the # sign... so if js is turned off then the urls would act as they normally do... all "apps" (each section of the site) has its own keyword... so all i need to do is add to the js array above to add more pages...
So for example my current settings are set to:
var objApps= ['ads','user'];
So if i have a url such as:
www.domain.com/ads/3923/bla/dada/bla
the js script would replace the /ads/ part so my URL would end up being
www.domain.com/ads/#p=3923/bla/dada/bla
Then I use jquery bbq plugin to load the page accordingly...
http://benalman.com/projects/jquery-bbq-plugin/

I have found the metadata plugin to be an excellent solution to the problem of storing arbitrary data with the html tag in a way that makes it easy to retrieve and use with jQuery.
Important: The actual file you include is is only 5 kb and not 37 kb (which is the size of the complete download package)
Here is an example of it being used to store values I use when generating a google analytics tracking event (note: data.label and data.value happen to be optional params)
$(function () {
$.each($(".ga-event"), function (index, value) {
$(value).click(function () {
var data = $(value).metadata();
if (data.label && data.value) {
_gaq.push(['_trackEvent', data.category, data.action, data.label, data.value]);
} else if (data.label) {
_gaq.push(['_trackEvent', data.category, data.action, data.label]);
} else {
_gaq.push(['_trackEvent', data.category, data.action]);
}
});
});
});
<input class="ga-event {category:'button', action:'click', label:'test', value:99}" type="button" value="Test"/>

My answer might not apply to your case. I needed to store a 2D table in HTML, and i needed to do with fewest possible keystrokes. Here's my data in HTML:
<span hidden id="my-data">
IMG,,LINK,,CAPTION
mypic.jpg,,khangssite.com,,Khang Le
funnypic.jpg,,samssite.com,,Smith, Sam
sadpic.png,,joyssite.com,,Joy Jones
sue.jpg,,suessite.com,,Sue Sneed
dog.jpg,,dogssite.com,,Brown Dog
cat.jpg,,catssite.com,,Black Cat
</span>
Explanation
It's hidden using hidden attribute. No CSS needed.
This is processed by Javascript. I use two split statements, first on newline, then on double-comma delimiter. That puts the whole thing into a 2D array.
I wanted to minimize typing. I didn't want to redundantly retype the fieldnames on every row (json/jso style), so i just put the fieldnames on the first row. That a visual key for the programmer, and also used by Javascript to know the fieldnames. I eliminated all braces, brackets, equals, parens, etc. End-of-line is record delimiter.
I use double-commas as delimiters. I figured no one would normally use double-commas for anything, and they're easy to type. Beware, programmer must enter a space for any empty cells, to prevent unintended double-commas. The programmer can easily use a different delimiter if they prefer, as long as they update the Javascript. You can use single-commas if you're sure there will be no embedded commas within a cell.
It's a span to ensure it takes up no room on the page.
Here's the Javascript:
// pull 2D text-data into array
let sRawData = document.querySelector("#my-data").innerHTML.trim();
// get headers from first row of data and load to array. Trim and split.
const headersEnd = sRawData.indexOf("\n");
const headers = sRawData.slice(0, headersEnd).trim().split(",,");
// load remaining rows to array. Trim and split.
const aRows = sRawData.slice(headersEnd).trim().split("\n");
// trim and split columns
const data = aRows.map((element) => {
return element.trim().split(",,");
});
Explanation:
JS uses lots of trims to get rid of any extra whitespace.