[jdom-interest] getChild and namespace

[elephantwalker]

Dear Philip,

This parses in crimson or xerces, latest release. Many parsers *glorck* on
svg because the svg dtd has *many* entity references. For example, the
oracle parser with and without jdom will not work...no matter what version
you use.

A good test for any parser is to parse svg xml. You can get several examples
from Adobe's svg site (www.adobe.com) or you can look at the w3c stuff at
www.w3c.org.

Regards,

Elephantwalker

-----Original Message-----
From: philip.nelson at omniresources.com
[mailto:philip.nelson at omniresources.com]
Sent: Tuesday, March 27, 2001 11:41 AM
To: elephantwalker at home.com; jdom-interest at jdom.org
Subject: RE: [jdom-interest] getChild and namespace



> I'm a little confused about something.
>
> If you have a default namespace (usually the name of the dtd
> without the
> .dtd extension), and there are other namespaces declared in
> the root element
> of your document....
>
> <?xml version="1.0" encoding="iso-8859-1"?>
> <!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 20000303 Stylable//EN"
> "http://www.w3.org/TR/2000/03/WD-SVG-20000303/DTD/svg-20000303
> -stylable.dtd"
> >
> <svg  xmlns:a="http://www.adobe.com/svg10-extensions"
> a:timeline="independent">
>   <defs>
>     <symbol id="rectangle" >
>     <path  d="M0,0L0,50L50,50L0,0z" />
>     </symbol>
>   </defs>
>   <g>
>     <use xlink:href="#rectangle" />
>   </g>
> </svg>
>
> If you do a doc.getRootElement(), and then get the Namespace
> it is the path
> to the dtd without the dtd extension.

I can't even parse this.  I get a StringIndexOutOfBounds exception from
xerces (ver 1.2).  I tried another svg example and the same thing happened.
what parser are you using?

When I try to create my own doctype in a similar fashion but without
including the entities that svg is including in the dtd, I end up with a
NO_NAMESPACE namespace so i don't think that the doctype declaration has
anything to do with a default namespace uri of "svg".  I have to think I've
completely missed the boat on your explanation but that wouldn't even be a
valid uri.

> If you do a
> root.getChild("defs"), the
> return is null, but if you do
> root.getChild("defs",root.getNamespace()), you
> get the symbol element.

What!!!  This is why I spent the time with this because this would be a
serious problem, no?

> If I have "glorcked" to much here, and this is just a bug...

I hope not too...