Re: Banning relative - No real damage?

> However, the practicaility of it that not one single instance of
> a document has been brought as evidence that this is a real
> problem.  Everyone pointed at Microsoft, and Microsoft produced an example
> http://lists.w3.org/Archives/Public/xml-uri/2000May/0145.html
> which does *not* in fact use relative URIs.

That's true but their new MSXSL 3 XSL system certainly does.
(I am not sure about the XSL dialect built into IE5)
MSXSL uses the XSLT extension namespace mechanism to extend XSLT and
all the examples in the SDK documentation use relative URL to
script implementations of the extension functions.
I am not sure they have documentation on line, and I don't have a
machine running windows in this building to find an example in the
documentation, but perhaps someone could point you to a reference.

Having said that, of two unpalatable alternatives, "forbid" is probably
better than "absolute". If it had been done at the start there would
have been no problem with forbid at all, it is functionally equivalent
to literal, the only problem with forbidding now is existing
documents. 

You'll find some examples on xsl-list using a namespace of "/dev/null"
the examples themselves are of course just toy examples, but
presumably the original poster was asking about a real question, so
the construct may by now have found its way into any number of real
documents, how are we to know?

> And for the Infoset we can define a behaviour if the banned form
> is met, of course.

literal comparison?

David

Received on Monday, 5 June 2000 18:46:42 UTC