I don't think we should use packages in org.relaxng for
implementations. They should be reserved for interfaces agreed on by
more than one implementor. Similarly, namespaces within
http://relaxng.org ought to be reserved for things that we work on as
a committee. If the TC decides to come up with some common additional
datatypes beyond those in XML Schema Part 2, then we can use a
namespace within http://relaxng.org. Until then, I think you should
choose a namespace in some other domain.
----Original Message-----
>From: Kohsuke KAWAGUCHI <>
>To:
>Cc:
>Subj: Datatype interface reference implementation
>Reply To:
>Sent: 20 ???????? 2544 14:27
>
>
>Here is the reference implementation of RELAX NG datatype
interface for
>Java.
>
>It uses the namespace URI of "http://relaxng.org/ns/xpathlib/1.0"
(is
>this OK?), and validates XPath.
>
>It is based on Ingo Macherius's XPath grammar for JavaCC. And the
>current release doesn't contain licensing terms. It is intended to
be
>released under the same license as the interface itself.
>
>
>
>> For example, the following example validates "path" attribute as
>> an XPath.
>>
>> <element name="test">
>> <attribute name="path"
>> datatypeLibrary="http://relaxng.org/ns/xpathlib/1.0">
>> <data type="XPath"/>
>> </attribute>
>> </element>
>>
>> It matches things like:
>>
>> <test path="//abc/def/ghi[@jkl='mno']"/>
>> <test path="child::chapter/descendant::para"/>
>>
>> But not
>>
>> <test path="foo:chapter//title"/>
>> --- undeclared namespace prefix "foo".
>> <test path=":::"/>
>> --- bad XPath
>
>
>regards,
>--
>Kohsuke KAWAGUCHI +1 650 786 0721
>Sun Microsystems
>
