[
Lists Home |
Date Index |
Thread Index
]
Julian Reschke wrote:
>>From: Emmanuil Batsis (Manos) [mailto:mbatsis@humanmarkup.org]
>>
>>
>>That's exactly the option I'm after: a one to one relationship between a
>>namespace and a schema, and a one to many relationship between a
>>
>>
>
>There isn't on in the general case. Consider XHTML.
>
Where did that came from? XHTML uses doctypes/DTDs and even XHTML m12n
in schema does not even come close to what we are talking about.
>>document and scemas (through namespaces), meaning mixed documents.
>>
>>
>
>In the general case, it's many-to-many.
>
>
>
>>>>Change your namespace URI to a URL. Do you think this presents
>>>>
>>>>
>>a problem?
>>
>>
>>>>
>>>>
>>>Yes. Because by doing that, you define a *new* vocabulary, and
>>>
>>>
>>this change
>>
>>
>>>will break processing of "old" documents.
>>>
>>>
>>>
>>How will processing of old documents break? Those will continue to work
>>as they where designed to. The new one will work as the old one used to,
>>with the addition of validation abilities.
>>
>>
>
>Changing the namespace name means that no existing code will process these
>documents, as it would be testing for the old namespace.
>
Nor should existing code process the new namespaced document. However
you haven't answered my question on how will processing of *old*
documents break.
>I'd expect that the default namespace for each of a document's elements is
>available through the namespace axis. Did you try that?
>
Heh. The namespace axis wont tell you it's the *default* namespace so
you don't know if you have already handled that for the contect node (or
one of it's ancestors).
Stuff like
<xsl:if test="not(ancestor::*[namespace-uri()=$nsUri]) and $nsUri != ''">
<xsl:apply-templates select="." mode="ns-resolve"/>
</xsl:if>
are irritating.
Manos
|