Reloading existing classes with existing data is likely to break things.
You can load new code into new class loaders relatively easily:
ClassLoader loader = URLClassLoader.newInstance(
new URL[] { yourURL },
getClass().getClassLoader()
);
Class<?> clazz = Class.forName("mypackage.MyClass", true, loader);
Class<? extends Runnable> runClass = clazz.asSubclass(Runnable.class);
// Avoid Class.newInstance, for it is evil.
Constructor<? extends Runnable> ctor = runClass.getConstructor();
Runnable doRun = ctor.newInstance();
doRun.run();
Class loaders no longer used can be garbage collected (unless there is a memory leak, as is often the case with using ThreadLocal, JDBC drivers, java.beans, etc).
If you want to keep the object data, then I suggest a persistence mechanism such as Serialisation, or whatever you are used to.
Of course debugging systems can do fancier things, but are more hacky and less reliable.
It is possible to add new classes into a class loader. For instance, using URLClassLoader.addURL. However, if a class fails to load (because, say, you haven't added it), then it will never load in that class loader instance.