CS360 Lecture Notes -- The Jval Type

  • James Plank
  • Directory: ~plank/cs360/notes/Jval
  • Lecture notes: http://web.eecs.utk.edu/~plank/plank/classes/cs360/360/notes/Jval
  • Tue Oct 13 10:07:44 EDT 1998

    The Jval type

    In jval.h, I define a type called a Jval. This is a big union:
    typedef union {
        int i;
        long l;
        float f;
        double d;
        void *v;
        char *s;
        char c;
        unsigned char uc;
        short sh;
        unsigned short ush;
        unsigned int ui;
        int iarray[2];
        float farray[2];
        char carray[8];
        unsigned char ucarray[8];
      } Jval;  
    I make use of Jval's when I write generic data structures such as lists, and trees. For the purposes of this class, don't worry about half of the fields in the union. The only important ones are:
        int i;
        float f;
        double d;
        void *v;
        char *s;
        char c;
    The nice thing about a Jval is that you can hold one piece of data in a Jval, regardless of what the type of that piece of data is. Moreover, the Jval will always be 8 bytes. You use it just like the unions that were discussed in the union lecture notes from CS140. Take a minute to go over that lecture to brush yourself up on what a union is.

    Constructor functions

    You can of course, create and use a Jval by simply declaring and using it. For example:
      Jval j;
      j.i = 4;
    You can freely pass Jval's to and from procedure calls. A Jval can be the return value of a procedure call.

    Jval.h defines a whole bunch of prototypes for ``constructor functions.''

    extern Jval new_jval_i(int);
    extern Jval new_jval_f(float);
    extern Jval new_jval_d(double);
    extern Jval new_jval_v(void *);
    extern Jval new_jval_s(char *);
    These return Jval's to you when you give them arguments of a specific type. For example, if you want to initialize a Jval so that it is in integer whose value is 4, you can do it as shown above, or you can do:
      Jval j;
      j = new_jval_i(4);
    Now j.i will be the integer 4. You will see later (when we get to using some of the general purpose data structures) why this is convenient.

    These ``constructor functions'' are implemented in jval.c, and are very simple. For example, here is new_jval_i():

    Jval new_jval_i(int i) {
      Jval j;
      j.i = i;
      return j;

    A simple example

    Obviously, to use Jval's, include jval.h, and then link with libfdr.a as described in the Libfdr lecture notes..

    As an example, jval_ex.c is union3.c from the Union lecture, converted to use Jvals.

    You'll note that there is only one change to the code: the typedef:

    typedef struct {
      char type;
      Jval value;
    } Item;
    Since the Jval struct has the same .i, .f and .s fields, the rest of the code can remain unchanged.

    As you can see, it works just fine:

    UNIX> jval_ex
    int 4
    string Jim
    float -33.2
    int -2
    int 1
    Item 0: Type i -- Value: 4
    Item 1: Type s -- Value: Jim
    Item 2: Type f -- Value: -33.200001
    Item 3: Type i -- Value: -2
    Item 4: Type i -- Value: 1
    Sizeof(Item): 16

    In jval_ex2.c, we modify the code to use the constructor functions. The main change is (and I've bold-faced the new code):

      int i2;
      float f;
      for (i = 0; i < 5; i++) {
        if (get_line(is) != 2) exit(1); 
        if (strcmp(is->fields[0], "int") == 0) {
          array[i].type = 'i';
          if (sscanf(is->fields[1], "%d", &i2) != 1) exit(1); 
          array[i].value = new_jval_i(i2);
        } else if (strcmp(is->fields[0], "float") == 0) {
          array[i].type = 'f';
          if (sscanf(is->fields[1], "%f", &f) != 1) exit(1); 
          array[i].value = new_jval_f(f);
        } else if (strcmp(is->fields[0], "string") == 0) {
          array[i].type = 's';
          array[i].value = new_jval_s(strdup(is->fields[1]));
        } else {

    A word of warning about Jval's

    The purpose of the Jval type is to make general purpose data structures such as dllists and red-black trees as flexible and efficient as possible. You are not to use Jval's in your code for any other reason. I will tell you when to use them.

    Specifically, you are not to say, use a Jval instead of an int in your code just because it works. That makes your code unreadable, and unreadability is majorly bad. Here is an example of bad code to average all of the integers on standard input (in badavg.c):

      Jval total;
      Jval j;
      Jval n;
      n.i = 0;
      total.i = 0;
      while (scanf("%d", &(j.i)) == 1) {
        total.i += j.i;
      total.d = ((double) total.i) / ((double) n.i);
      printf("Average = %lf\n", total.d);
    Yes, it works, and yes, it's a cute way to use total as both an int and a double. But it is revolting -- every use of Jval's is bad, and if you use them in ways like these, you will be punished.

    (In case you care, the code should look as in goodavg.c):

      int total;
      int j;
      int n;
      n = 0;
      total = 0;
      while (scanf("%d", &j) == 1) {
        total += j;
      if (n == 0) exit(1);
      printf("Average = %lf\n", ((double) total)/((double) n));
    (You can do other things to make that casting look better too).

    Accessor functions

    I have put accessor functions into jval.h/jval.c. An accessor function simply lets you get the desired value out of a Jval by calling a function rather than accessing the field. Why would you want to do this? Well, like the constructor functions, it makes life easier in certain circumstances. The accessor functions are:
    extern int    jval_i(Jval);
    extern long   jval_l(Jval);
    extern float  jval_f(Jval);
    extern double jval_d(Jval);
    extern void  *jval_v(Jval);
    extern char  *jval_s(Jval);
    extern char   jval_c(Jval);
    So, for example, calling jval_i(j) is the same as using j.i.


    Finally, jval.h contains a global variable JNULL. Use this when you would use NULL for a char * or void *.