Re: [xsl] random url for namespace to avoid uncontrollable dependencies

[Andrew Welch <andrew.j.welch@xxxxxxxxx>]

On 25 August 2011 15:34, David Carlisle <davidc@xxxxxxxxx> wrote:
> On 25/08/2011 15:01, Andrew Welch wrote:
>>
>> where does any resolution to an absolute uri take place?
>
> It doesn't.
>
> but some people (who I could name:-) would have wished that namespaces
> were defined differently and that you could rely on resolving the
> namespace URI to something. (You see the same tension in XSD schema spec
> suggesting that perhaps or perhaps not, depending on the phase of the
> moon, you might, or possibly might not, find a schema document at the
> namespace URI).
>
> So after (literally) a few thousand acrimonious exchanges in the
> xml-uri mailing list the thing went to a ballot of wc xml plenary with the
> end result as quoted earlier. Essentially, that as people couldn't agree
> on what relative namespace uri meant, and that some people even if they
> did agree, would want to change the spec to make it say something else,
> that the safe course of action was to just tell people not to use
> relative URI, so that they were future proofed against any future
> "re-interpretation" of the spec.

Interesting... that's the situation now - if it's a url you may find
the xsd, an html page, something else or nothing at all... which is no
more helpful than allowing any value as the namespace name.

If at all that happens is an html page exists at the url (as per
http://www.w3.org/1999/XSL/Transform), then that's no better (actually
its worse) than adding an attribute to the root element:

<foo xmlns="whatever-i-like" markup-description="http://www.mycomp.com/xml-v1/";>

At least then if nothing is at the url, you know there is a problem.
And the url can change between versions while the namespace name
remains the same...


-- 
Andrew Welch
http://andrewjwelch.com