[
Lists Home |
Date Index |
Thread Index
]
> From: Elliotte Rusty Harold [mailto:elharo@metalab.unc.edu]
> Sent: Saturday, September 28, 2002 2:45 PM
> To: Julian Reschke; xml-dev@lists.xml.org
> Subject: Re: [xml-dev] namespaces in XML 1.1 - namespace names become
> *IRI* references
>
>
> At 2:13 PM +0200 9/28/02, Julian Reschke wrote:
>
>
> >Namespace names allowing characters outside the ASCII character set means
> >that this change will break any code that currently relies on
> the fact that
> >namespace names must be valid URI references according to
> RFC2396. Of course
> >code can be changed, but I doubt that the cost of this has been
> considered
> >properly.
> >
>
> Namespaces 1.1 will only be available with XML 1.1. XML 1.1 has much
> more serious (and much more breaking) changes than this. Existing XML
> 1.0 parsers will generally not be able to read XML 1.1 documents.
> Among other things, the definition of name characters and white space
> is likely to change.
Yes, I agree that the transition from XML 1.0 to XML 1.1 is more serious for
*XML processors*. However, I'm talking about *applications*. Applications
will not be concerned at all about changes in whitespace handling, nor will
the extension of valid name characters likely cause any problems (it's just
that "more" Unicode characters are allowed). On the other hand, allowing
non-ASCII characters in namespace names *is* going to create problems,
because there is deployed code which relies on URI syntax for namespace
names.
As the switch to IRIs doesn't seem to be motivated by the requirements
document, I'd at least like to see the *rational* for this incompatible
change (there *is* a requirement to keep the changes minimal: "The changes
required for Namespaces in XML 1.0 processors to also process Namespaces in
XML 1.1 must be as few and as small as possible.").
Julian
--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
|
