Making Software

An Example of Code Cloning

Before we go further, we should probably explore the problem space a little. To that end, let’s consider the following two functions that are taken from version 1.6.3 of the Gnumeric source code base:

// Both functions are from the file py-gnumeric.c, version 1.6.3
static PyObject *
py_new_Range_object (GnmRange const *range) {
        py_Range_object *self;
        self = PyObject_NEW (py_Range_object, &py_Range_object_type);
        if (self == NULL) {
                return NULL;
        }
        self->range = *range;
        return (PyObject *) self;
}

static PyObject *
py_new_RangeRef_object (const GnmRangeRef *range_ref){
        py_RangeRef_object *self;
        self = PyObject_NEW py_RangeRef_object, &py_RangeRef_object_type);
        if (self == NULL) {
                return NULL;
        }
        self->range_ref = *range_ref;
        return (PyObject *) self;
}

At first glance, it seems pretty obvious that one of these functions has been cloned from the other (or that both have been cloned from a third source), as they are almost token-for-token identical. Indeed, a little investigation reveals that these are boilerplate designs for dealing with the instantiation of Python objects from within the main C codebase. The functions differ in two ways, one obvious and one more subtle. The obvious difference is that some of the identifiers are slight variations on each other: each identifier that has the suffix Range in the first function has the suffix RangeRef in the second. The more subtle difference concerns the parameters: in the first function, the parameter is a constant ...