Non-(X)HTML Attributes... any disadvantages?

Question

I've generally tried to stick with DOM-only attributes when writing Javascript. Now that I've switched from Prototype to jQuery, I can get some serious mileage out of adding my own attributes to various DOM elements, mostly in the realm of being able to set up a very readable coding convention for handling AJAX requests.

As a short example, this means I do things like

<div type="book" app_id="13">
    <a href="#" action="delete">delete</a>
</div>

And then I can set up code to find all <a> tags with an action attribute, find a parent with a type and app_id, and then do CRUD operations... all without me having to write additional code.

Are there any pitfalls (other than not being strictly XHTML complaint) that I should watch out for, and/or any good habits I should look to emulate? How about a standard way of setting up my own attribute namespace?

Answer 1

According to this question, using XML namespaces in XHTML 1.0 is invalid. Adding your own attributes to the same namespace seems worse to me, as they're most certainly invalid, even as far as XML goes.

Were I doing this, I'd get my mileage out of class and rel attributes. For example:

<div class="book" id="book_13">
   <a href="http://example.com/url/to/delete/non/ajaxily" class="delete">delete</a>
</div>