Hmmmm. There's a small chance that this patch will also trigger the old bug 
#1006, which was that file access is not allowed in applets. Accessing .jar 
files should be OK, but you never know. So that would need to be tested. 
(Basically, just build the Vizlet, then try it out, taking great care to flush 
the browser cache thoroughly first.)

This method gets called from all over Ontopia, and so I worry a little that it 
may have a negative performance impact. Whether it really does I'm not sure.

Finally, I wonder what the use of the patch is. Where/when would we want to 
make URLs to resources on the classpath? (Note that I'm not saying we wouldn't 
do it. I'm just wondering what the benefit is.)

The intended use of this patch is being able to use e.g. 
ImportExportUtils.getReader(String) with a classpath reference. Other methods 
might be adapted later to load files from classpath as well.

To avoid (most of the) overhead, and to be more explicit, I've updated the 
patch. It now checks if the argument starts with a "classpath:" prefix. That 
way it's both clear what the intention and result is and it makes less impact 
on performance. Also, StreamUtils.getResource will give an explicit error if 
the resource was not found on classpath.

This new patch will probably lower or even eliminate chances of triggering bug 
#1006. I haven't tested in yet. I will do that later and commit this new patch 
if no further comments or objections arise.

I've updated the patch again and deleted the previous. It throws an 
OntopiaRuntimeException if the classpath resource does not exist.

I've applied the proposed patch to the Maven branch in revision r1615. We're 
using it to reach test files from classpath references.

Maven branch has been merged to trunk, so proposed fix is now in trunk.