The problems with namespaces in text/html (Was: MathML-in-HTML5)

On Thu, 5 Oct 2006 juanrgonzaleza@canonicalscience.com wrote:
>
> Ian Hickson said:
> >
> > There are literally tens of millions of pages, for instance, that use 
> > the "xmlns" attribute on the <a> element,
> 
> What is the problem with using the xmlns attribute on <a> or <html:a> 
> elements in a XML approach?

In browsers today, the following:

   <a href="test" xmlns=""> ... </a>

...is just a link. If we start supporting xmlns="" as it works in XML, but 
in HTML, then literally millions of pages are going to suddenly have their 
links stop working, because <a> in the "" namespace (as opposed to the 
XHTML namespace), is not an HTML <a>, and thus isn't a link.

I've seen hundreds of thousands of occurances of bogus meaningless things 
like this:

   <br xmlns="http://www.w3.org/TR/REC-html40">

...as well as many thousands of pages with xmlns="" values pointing to 
their own sites (as opposed to any sort of half-sensible namespace).

Some pages even have completely bogus namespaces on the root <html> 
element, which would make the entire page screw up. Even worse, Office 
HTML, of which there is a LOT on the Web, uses namespaces in a way to 
trigger IE to do one thing, but relies on the other browsers *not* 
handling the namespaces to make sure it all works everywhere. (Like I said 
earlier, I've worked with one browser vendor who tried implementing this 
namespace thing before, and had to back out because it broke real content 
in pretty fundamental ways.)

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Thursday, 5 October 2006 19:41:04 UTC