This is a SUNBUG for https://bugs.openjdk.java.net/show_bug.cgi?id=100129

Description From Guillaume C:

Created an attachment (id=174) [details]
demo of the problem

When scripting through Rhino (e.g. selecting the "JavaScript" script engine),
whenever an exception is thrown by a Java class (called by javascript), it is
not exposed by the thrown ScriptException. Here's a demo:

-=-=---=-=---=-=--
import java.util.Hashtable;
import javax.script.*;

public class Test {

    private static class BuggyClass {
        private Hashtable store = new Hashtable();
        public BuggyClass() {
            System.out.println("Creating BuggyClass instance...");
            store.put(null, null); // BOOM
        }
    }

    public static void main(String[] args) {
        ScriptEngineManager manager = new ScriptEngineManager();
        ScriptEngine engine = manager.getEngineByName( "JavaScript" );
        try {
            engine.eval("new Packages.Test.BuggyClass();");
        } catch ( Exception e ) {
            Throwable t = (Throwable)e;
            while (t != null) {
                System.err.println("Exception: " + t.getClass());
                t = t.getCause();
            }
        }
    }
}
-=-=---=-=---=-=--

This outputs:

Creating BuggyClass instance...
Exception: class javax.script.ScriptException

The ScriptException object lacks a cause which would allow to obtain the
underlying NullPointerException (with its stacktrace), located in the
BuggyClass implementation; for deeper classes hierarchy than in this demo, it
makes it nearly impossible to debug/locate where is the problem, although
normally the "#getCause" of Throwable object makes it very efficient.

With that modification to
src/share/classes/com/sun/script/javascript/RhinoScriptEngine.java the cause
would be present:

-=-=---=-=---=-=--
--- RhinoScriptEngine.java 2010-02-19 09:47:37.000000000 +0100
+++ RhinoScriptEngine.java.patched 2010-02-19 17:45:00.000000000 +0100
@@ -107,7 +107,9 @@
         } catch (RhinoException re) {
             if (DEBUG) re.printStackTrace();
             int line = (line = re.lineNumber()) == 0 ? -1 : line;
- throw new ScriptException(re.toString(), re.sourceName(), line);
+ ScriptException se = new ScriptException(re.toString(),
re.sourceName(), line);
+ se.initCause(re);
+ throw se;
         } catch (IOException ee) {
             throw new ScriptException(ee);
         } finally {
@@ -181,7 +183,9 @@
         } catch (RhinoException re) {
             if (DEBUG) re.printStackTrace();
             int line = (line = re.lineNumber()) == 0 ? -1 : line;
- throw new ScriptException(re.toString(), re.sourceName(), line);
+ ScriptException se = new ScriptException(re.toString(),
re.sourceName(), line);
+ se.initCause(re);
+ throw se;
         } finally {
             cx.exit();
         }
-=-=---=-=---=-=--

And the above example program would output:

Creating BuggyClass instance...
Exception: class javax.script.ScriptException
Exception: class org.mozilla.javascript.WrappedException
Exception: class java.lang.NullPointerException

making is much more useful!

This was checked against lastest stable, 1.6.0_18.

Comment #1 From Guillaume C:

Created an attachment (id=175) [details]
proposed patch to RhinoScriptEngine.java