Re: xml:id versus xmlid

Henry S. Thompson wrote:

> Having maintained and developed a large Java XML API for some years, I
> agree that the xml namespace and prefix require special casing.  That
> special casing took work.  That work is done.  The marginal cost of
> supporting xml:id given that the work has been done is tiny.

If I were to take the time to produce a list of existing APIs and tools 
in which that work has not been done, would this convince you to support 
xmlid instead of xml:id?

> The conceptual overhead of explaining that global attributes are
> namespace-prefixed, except when they're spelled 'x m l i d', seems to
> me _much_ more costly when averaged over all likely users, than the
> cost of getting xml: right when averaged over all developers.

The conceptual overhead is explaining global attributes, full-stop. 
Irrespective of any xml: issues, this is one of the things in the W3C 
Schema language that is consistently confusing and regularly befuddles 
students. The proper solution for xmlid is simply not to mention or 
claim any distinction between global and local attributes. The concept 
is unnecessary and not useful. It is much easier to explain that xmlid 
and xml:space and xml:lang are reserved attributes that have particular 
semantics. That's all.

-- 
Elliotte Rusty Harold  elharo@metalab.unc.edu
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim