Re: load on DocumentLS vs framework load
Curt Arnold <[email protected]>
| Newsgroups | gmane.comp.web.dom.test-suites.general |
|---|---|
| Message-ID | <[email protected]> |
Jeroen van Rotterdam wrote:
> I believe there's a naming conflict between the load element of the test
> suite and the load method of DocumentLS in DOM L3 Load & Save.
>
>
> This is what my generated dtd says about load:
>
> <!ELEMENT load EMPTY>
> <!ATTLIST load
> var CDATA #REQUIRED
> href CDATA #IMPLIED
> uri CDATA #IMPLIED
> obj CDATA #IMPLIED
> interface (DocumentLS) #IMPLIED
> willBeModified (true | false) #REQUIRED
>
>
>
> And here's a part of my test:
>
> <assign var="documentls" value="document"/>
> <load var="result" obj="documentls" uri="TEST1" willBeModified="false"
> interface="DocumentLS"/>
>
>
> Which has an attribute willBeModified which is added because it is
> required in the dtd. It should be eliminated.
>
> Here's the java code generated by the stylesheet:
>
>
> documentls = (DocumentLS) document;
> result = load("");
>
>
> I assume the stylesheet is looking at the href since it assumes this a
> load by the framework instead of DOM L3-LS.
>
>
> Any idea ?
I committed two small changes to test-to-java.xsl and dom-to-dtd.xsl.
In test-to-java, I changed the test on the template that produces the
framework load from:
<xsl:template match="*[local-name() = 'load']" mode="body">
to
<xsl:template match="*[local-name() = 'load' and not(@interface)]"
mode="body">
That should let an DocumentLS load fall through to the DOM method
generation code.
In dom-to-dtd.xsl, I made willBeModified optional to be consistent with
the schema.
Please let me know if these adequately address the issue.