Lists Home |
Date Index |
Michael Kay wrote:
> James Anderson asked:
> > Are there really cases where one copies text nodes from a context in
> > which a prefix had one binding to a context in which a prefix has a
> > different binding with the intent of effecting a change in the implied
> > universal names.
> It may be that the prefix has a different binding in the target context;
> it's more likely that it has no binding.
Which means that, while the special case does work, for all the effort,
in the general case, the product can be an mal-formed dom. Even where is
it expected that one is testing for duplicate attribute names, that just
means that the operation will faill unnecessarily.
> Consider a stylesheet:
> <xsl:stylesheet xmlns:xsl="...." version="1.0"
> <xsl:template match="circle">
> <area><xsl:value-of select="math:pi() * @radius * @radius"/></area>
> You now want to copy this template rule to a different stylesheet.
Yes, I would.
> To do so,
> you have to copy the namespace declaration xmlns:math,
No you don't. If you model the value of the select attribute in the
> otherwise the
> template rule is meaningless. And of course, you want to do the copy using
> general-purpose XML tools, tools which don't understand that there's
> anything special about the select attribute of an <xsl:value-of> element.
We differ here. To pose a rhetorical question, are they really xml
tools? It's not an issue of "allow". it's a question of whether one
should ever have expected it to work.
> Now, I hear lots of people saying we shouldn't have allowed this, and I'm
> inclined to agree. But we do allow it, and it works today, and people are
> taking advantage of the fact that it works today.
Years ago, when I worked as an operator for BBN, there were occasions
when I would witness TENEX wizards hunched over the operator's console
typing system patches into the debugger. I was in school then, so I
worked nights and weekends. Which means I never had a chance to overhear
the conversations between the CEO and one of them when they made a typo.
Maybe they never did. I suppose that means the process worked.
The computer room was pretty big. Eventually a twin appeared next to the
production system and it was more likely to see a system wizard hunched
over its console.
> And I have yet to see a
> proposal that cleans up the namespace model without breaking applications
> that work today according to the current specs.
To the extent that an application expects a prefix-namespace binding
which was apparent in the dynamic context of a given document's parser
to have indefinite extent, the application is already broken.