Question

I have been trying to generate some dynamic codes (Using Javassist) but program fails at a certain point when involving a double array or float array. The code is as follows

Class c = Customers.class; // called in main & Customer class just has a double[] Dubs = new double[10]
CreateType(c); // Main


public static Object CreateType(Class genericType)
{
        // some preReq declarations
        CtMethod writeCode = dyn.getDeclaredMethod("processCode");

        generateCode(genericType, Code, "temp"); // Code is a StringBuilder class

        System.out.println(Code);

        writeCode.insertAt(1, Code.toString()); // Compilation is successful

        Class c = dyn.toClass();

        Dynamic h;
        Constructor[] ctorlist = null;

        ctorlist =  c.getDeclaredConstructors(); // Problem is here

        h = (DynamicSurrogate) ctorlist[0].newInstance(genericType);

        return h;
}

Generated code is as follows

    testapp1.Customers temp=(testapp1.Customers)graph;
    output.processDouble(temp.Dubs[1]);

But problem arises when getDeclaredConstructors is called c.getDeclaredConstructors() … it throws the following error

Exception in thread “main” java.lang.VerifyError: (class: testapp1/Dyn, method: processDouble signature: (Lsomething/Output;Ljava/lang/Object;)V) Inconsistent args_size for opc_invokeinterface

A workaround exists but does not make any sense, i.e. everything works fine if i simply create a copy of the double array and pass it on to processDouble in dynamic code
i.e. if the dynamic code is

    testapp1.Customers temp=(testapp1.Customers)graph;
    double[] d = temp.Dubs;
    output.processDouble(d);

In short , exception Unhandled is thrown by getDeclaredConstructor but it actually has nothing to do with a constructor because it doesnt matter if i create one or not

Hopefully my problem and code is clear enough, if any confusion please do tell, Thankyou in advance 🙂

Answer 1

The verify error is thrown out of getDeclaredConstructors because that’s the first place that the bytecodes in class c are verified.

As to the cause of the error, I suspect it has to do with the fact that in the first case you’re passing a double, and in the second case a double[].