Re: [XML-URI] HTTP extensions framework comparison

> At 09:01 20/07/00 -0700, Henrik Frystyk Nielsen wrote:
> >Very similarly, an application that only does case-sensitive comparison
of
> >the full http URI will not know that
> >"http://www.informatix.co.uk/headers/" and
> >"http://WWW.INFORMATIX.CO.UK/headers/" are equal, but the result is
> >exactly the same as if it didn't know the extension in the first place.
> >The important thing to note is that the stupid app *does* interoperate
as
> >it knows that it received an XML namespace and it knows that it doesn't
> >know about it.
> >
> >This of course applies directly to the XML NS debate.
>
> You are saying, then, that the XPath REC should permit, _but not
require_,
> XPath implementations (including XSLT) to normalise NSURIs in a scheme
> dependent manner when comparing names?

You can not require them to do that just as well as you can not require
them to know about a particular namespace.

> This can clearly result in a given XML source/XSLT stylesheet
combination
> producing different results depending on the normalisations performed by
> the particular XPath implementation.

If you don't want that possibility then don't use a URI space that has
this behavior - it is up to you, see [1].

> I beg to differ.  I believe this to be behaviour which should be
> standardised, and not left implementation defined.

It is - the URI spec already defines the normalization rules for the most
commonly used parts of current URIs and if so you care about this then you
should implement the URI spec or pick your URI space appropriately so you
don't run into this problem. Let me point out again that we can not change
the rules in each and every spec that uses URI - that would lead to true
lack of interoperabilility.

Henrik

[1] http://lists.w3.org/Archives/Public/xml-uri/2000Jun/0697.html

Received on Thursday, 20 July 2000 13:21:08 UTC