Simply
is sufficient.
The other forms are there for compatibility with XHTML; to make it possible to write the same code as XHTML, and have it also work as HTML. Some systems that generate HTML may be based on XML generators, and thus do not have the ability to output just a bare
tag; if you’re using such a system, it’s fine to use
, it’s just not necessary if you don’t need to do it.
Very few people actually use XHTML, however. You need to serve your content as application/xhtml+xml for it to be interpreted as XHTML, and that will not work in old versions of IE – it will also mean that any small error you make will prevent your page from being displayed in browsers that do support XHTML. So, most of what looks like XHTML on the web is actually being served, and interpreted, as HTML. See Serving XHTML as text/html Considered Harmful for some more information.
I think this quote from the HTML 5 Reference Draft provides the answer:
3.2.2.2 Void Elements
The term void elements is used to designate elements that must be empty. These requirements only apply to the HTML syntax. In XHTML, all such elements are treated as normal elements, but must be marked up as empty elements.
These elements are forbidden from containing any content at all. In HTML, these elements have a start tag only. The self-closing tag syntax may be used. The end tag must be omitted because the element is automatically closed by the parser.
HTML Example:
A void element in the HTML syntax. This is not permitted in the XHTML syntax.
Example:
A void element using the HTML- and XHTML-compatible self-closing tag syntax.
XHTML Example:
A void element using the XHTML-only syntax with an explicit end tag. This is not permitted for void elements in the HTML syntax.