Re: [xml-dev] How to get XPath in a XSLT shylesheet

["Andrew Welch" <andrew.j.welch@gmail.com>]

On 1/9/07, Michael Kay <mike@saxonica.com> wrote:
> > But I am interested in how many other vocabularies are
> > regularly used with more than one typical prefix. I guess
> > html is the leading contender (no prefix in xhtml, but prefix
> > in other places).
>
> The most common case, of course, is where a namespace is sometimes bound to
> a prefix and sometimes used as the default namespace.

Yes, I regularly come across root elements like this:

<foo xmlns="aaa" xmlns:foo="aaa">
  <foo:bar>...

I think it's because authors don't like (or didn't think it was
possible) to define the prefix on the element that uses the prefix,
eg:

<foo:foo xmlns:foo="aaa">

The technique which has won the prize for being most annoying is
defining the default namespace in the DTD as a defaulted attribute...

For example you look at the XML and see <document> but once its parsed
it's really <document xmlns="aaa">

When examining this kind XML for the first time I have to run an
identity transform just to get the expanded XML, and then use that as
basis for writing the transforms.

cheers
andrew