Generating invalid HTML for purpose of custom attributes

[kstubs]

Is it bad, or does it make parsing objects unstable if you append custom attributes to an HTML tag?  Lets say I want to keep track of a number, maybe a customers ID, so I do something like:

var div = new Element('div', {'customerID':1234}); 

which should result in:

<div customerID="1234"></div>

Thoughts?

[Walter Lee Davis]

HTML5 lets you do this, and pretty much anything else you like, by
adding a data- prefix to the attribute name. Have at it.

Walter

> --
> You received this message because you are subscribed to the Google
> Groups "Prototype & script.aculo.us" group.
> To post to this group, send email to prototype-s...@googlegroups.com
> .
> To unsubscribe from this group, send email to prototype-scripta...@googlegroups.com
> .
> For more options, visit this group at http://groups.google.com/group/prototype-scriptaculous?hl=en
> .

[kstubs]

Data prefix?

[T.J. Crowder]

On Mar 27, 4:41 am, kstubs <kst...@gmail.com> wrote:
> Data prefix?

Yes. Literally:

<div data-customer='1234'>

As of HTML5, any attribute whose name starts with "data-" is valid on
any element; they're ignored by browsers (other than being put on the
elements in the DOM so you can access them) and other tools:
http://www.w3.org/TR/html5/elements.html#embedding-custom-non-visible-data-with-the-data-attributes

It was added because people were doing exactly what you describe,
resulting in invalid documents and the potential for issues where
little-known attributes were inadvertently used, etc.

So for your requirement, you might want a `data-customer` attribute.

HTH,
--
T.J. Crowder
Independent Software Engineer
tj / crowder software / com
www / crowder software / com

[kstubs]

Perfect, thanks!

[RobG]

On Mar 27, 11:43 am, Walter Lee Davis <wa...@wdstudio.com> wrote:
> On Mar 26, 2011, at 9:37 PM, kstubs wrote:
> > Is it bad, or does it make parsing objects unstable if you append
> > custom attributes to an HTML tag? Lets say I want to keep track of
> > a number, maybe a customers ID, so I do something like:
>
> > var div = new Element('div', {'customerID':1234});

The issues are inadvertent overwriting of HTML attributes (so you
can't just use any attribute name, you have to be careful) and IE's
mishandling of DOM element attributes and properties.

To get consistency across browsers, you have to read the attributes
using getAttribute and set them (using code) with setAttribute.
Because IE munges attributes and properties, you should only ever use
DOM properties for HTML atributes.

So you need to be careful to distinguish between the two and only use
the appropriate method, which is why it is usually suggested to not
use custom attributes and to use a data object instead, that way you
only ever use one method that is consistent for all browsers.

> > which should result in:
> > <div customerID="1234"></div>

"Should" being the operative word. Note that in IE, the div DOM
element will have a property of customerID, but it will not in
Firefox. That sort of inconsistency is why you should avoid custom
attributes and properties.

Perhaps that issue is fixed in IE 9, but it will be a very long time
before you can ignore all other versions of IE on the web.

> HTML5 lets you do this, and pretty much anything else you like, by  
> adding a data- prefix to the attribute name. Have at it.

HTML5 is not a standard, nor is it widely supported yet.


--
Rob

[T.J. Crowder]

> Because IE munges attributes and properties, you should only ever use
> DOM properties for HTML atributes.

This is (as I understand it) one part of the rationale for the `data-`
prefix: There aren't any DOM element properties with those names, and
so IE's broken behavior isn't an issue with them. Yes, it does dump
them on the element instance (if you put a "data-foo" attribute on a
div, the element instance for that div will indeed have a property
called "data-foo" on it -- prior to IE9), but it's harmless (though as
always, YMMV).

For instance, try this:
http://jsbin.com/ewade3/2

That works fine on IE6, IE7, (I don't have IE8 handy), IE9, Chrome 10,
Safari 5, Firefox 3.6, and Opera 11 under Windows; and Chrome 10,
Firefox 3.6, and Opera 11 under Linux (Ubuntu 10.04 LTS). I don't have
a Mac handy, but it works in Mobile Safari on my iPhone. :-) The "IE
check"s show that IE6 and IE7 and presumably IE8 (but not IE9, yay) do
dump the "data-" properties on the element, but you wouldn't look for
them there anyway since no one else does -- stick to `getAttribute`
(or better yet, Prototype's `readAttribute` because of the *other*
insane things IE does with attributes) and you're fine.

> > HTML5 lets you do this, and pretty much anything else you like, by  
> > adding a data- prefix to the attribute name. Have at it.
>
> HTML5 is not a standard, nor is it widely supported yet.

True. But there are two very different aspects to HTML5: Codifying and
standardizing the things browsers were already doing and had been
doing forever, and defining new things for them to do. By its very
nature, the first part is widely supported. :-) "data-" attributes
fall into that category (every browser I've ever seen supported custom
attributes on elements; HTML5 reins it in a bit). I dare say that that
subset of HTML5 is a better specification for HTML in the real world
than the HTML4.01 standard from over 11 years ago. Of course, the
trick with the HTML5 spec is knowing which bits are which. ;-)

-- T.J.

[Moo]

Hi,

I created a little helper for custom data-attributes:
http://mstuhlmann.wordpress.com/2011/01/10/working-with-custom-data-attributes/

Greets,
mstuhlmann

[Victor]

Sorry for late answer, but there are my few coins on custom attributes.

For XHTML solution exists many years - from the beginning of XHTML itself: XML namespaces. Define custom attributes in custom namespace, and the resulting XHTML page will be perfectly valid. Example (stripped from working system):

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xmlns:a="Ajaxify framework" xmlns:v="Form navigation and validation framework">
<head>
  <title>Title</title>
</head>
<body>
  <div id="content"></div>
  <a href="content.html" a:target="content" a:confirm="Update content?">Refresh</a>
</body>
</html>


Works in almost any browser (IE6+, Opera 9+, Firefox 3.5+, Chrome 10+, Safari 4+ - each and every browser currently used by our developers, testers and clients), passes W3C validation. The biggest problem was escaping/not escaping semicolon in CSS selectors for different selector engines: Prototype 1.6 legacy selector and Prototype 1.7 + NWMatcher allows unescaped semicolon like $$("a[a:target]"), while Prototype 1.7 + Sizzle requires backslash $$("a[a\:target]").