In [3,4,5,6] there are sections in which we compare the methods implemented in our compiler with the above mentioned related works. However, we would like to state here the main differences between the work we presented in [5,6] and the Sagiv et all work , because, from our point of view, the Sagiv's group has reach the most relevant results in the area of shape analysis. The first main difference is that in our work we consider important properties, like the ``reference patterns'' and ``touch information'' among others (described next) which lead to a more precise description of the data structure used in the code. Second, we join similar reference shape graphs (RSG) to build a reduced set of RSG for each program point, while in  they keep all the graphs. We think that this may explain why their Three-Valued-Logic Analyzer (TVLA) run out of memory for simple codes like the singly linked list insert sort and bubble sort using the multiple structure approach . And third, they recognize that their TVLA engine is only useful to analyze small programs and they have report experimental results for singly linked list small operations (insert, reverse, sort, etc). However they have not published experimental results successfully dealing with real C codes based on combination of complex data structures like doubly linked list pointing to trees or to other lists, etc. Summarizing, and to the best of our knowledge, our compiler is the only one able to accurately identify the data structure at each sentence of a real C code. The analyzed codes are based on complex data structures such as doubly linked lists, trees, and octrees among others, and combinations of them, such as a doubly linked list of pointers to trees where the leaves point to doubly linked lists, etc.