Ctutor chapter 11

                      Chapter 11 - Structures and Unions


                            WHAT IS A STRUCTURE?

             A structure is a user defined data type.   You have the

        ability  to  define  a new type of  data  considerably  more

        complex than the types we have been using.  A structure is a

        combination  of  several different previously  defined  data

        types,  including other structures we have defined.  An easy

        to  understand definition is,  a structure is a grouping  of

        related  data in a way convenient to the programmer or  user

        of the program.   The best way to understand a structure  is

        to  look  at  an example,  so if you will load  and  display

        STRUCT1.C, we will do just that.

             The  program begins with a structure  definition.   The

        key  word  "struct"  is followed by  some  simple  variables

        between  the  braces,   which  are  the  components  of  the

        structure.   After  the  closing brace,  you will  find  two

        variables listed,  namely "boy",  and "girl".   According to

        the  definition  of a structure,  "boy" is  now  a  variable

        composed of three elements,  "initial",  "age", and "grade". 

        Each of the three fields are associated with "boy", and each

        can  store a variable of its respective type.   The variable

        "girl"  is also a variable containing three fields with  the

        same  names  as those of "boy" but  are  actually  different

        variables.  We have therefore defined 6 simple variables.

                         A SINGLE COMPOUND VARIABLE

             Lets  examine  the  variable "boy"  more  closely.   As

        stated above, each of the three elements of "boy" are simple

        variables  and can be used anywhere in a C program  where  a

        variable of their type can be used.   For example, the "age"

        element  is  an integer variable and can therefore  be  used

        anywhere  in a C program where it is legal to use an integer

        variable,  in calculations, as a counter, in I/O operations,

        etc.   The  only problem we have is defining how to use  the

        simple  variable  "age"  which is a  part  of  the  compound

        variable  "boy".   We  use both names with a  decimal  point

        between  them with the major name first.  Thus "boy.age"  is

        the  complete  variable name for the "age" field  of  "boy". 

        This  construct can be used anywhere in a C program that  it

        is desired to refer to this field.   In fact,  it is illegal

        to  use the name "boy" or "age" alone because they are  only

        partial definitions of the complete field.  Alone, the names

        refer to nothing.

                     ASSIGNING VALUES TO THE VARIABLES

             Using  the above definition,  we can assign a value  to

        each  of  the  three fields of "boy" and each of  the  three

        fields  of  "girl".   Note carefully that  "boy.initial"  is



                                  Page 73









                      Chapter 11 - Structures and Unions


        actually  a "char" type variable,  because it  was  assigned

        that in the structure, so it must be assigned a character of

        data.   Notice  that "boy.initial" is assigned the character

        'R'  in agreement with the above rules.   The remaining  two

        fields of "boy" are assigned values in accordance with their

        respective  types.   Finally  the three fields of  girl  are

        assigned values but in a different order to illustrate  that

        the order of assignment is not critical.

                     HOW DO WE USE THE RESULTING DATA?

             Now  that  we  have assigned values to the  six  simple

        variables, we can do anything we desire with them.  In order

        to keep this first example simple,  we will simply print out

        the values to see if they really do exist as  assigned.   If

        you carefully inspect the "printf" statements,  you will see

        that there is nothing special about them.  The compound name

        of each variable is specified because that is the only valid

        name by which we can refer to these variables.

             Structures  are  a very useful method of grouping  data

        together  in  order to make a program easier  to  write  and

        understand.   This  first example is too simple to give  you

        even  a hint of the value of using structures,  but continue

        on  through  these lessons and eventually you will  see  the

        value of using structures.

             Compile and run STRUCT1.C and observe the output.

                           AN ARRAY OF STRUCTURES

             Load  and  display the next  program  named  STRUCT2.C. 

        This  program  contains  the same  structure  definition  as

        before  but  this  time we define an array of  12  variables

        named "kids".   This program therefore contains 12 times 3 =

        36  simple variables,  each of which can store one  item  of

        data  provided  that  it is of the correct  type.   We  also

        define  a simple variable named "index" for use in  the  for

        loops.

             In order to assign each of the fields a value, we use a

        for loop and each pass through the loop results in assigning

        a  value to three of the fields.   One pass through the loop

        assigns all of the values for one of the "kids".  This would

        not be a very useful way to assign data in a real situation,

        but  a loop could read the data in from a file and store  it

        in  the correct fields.   You might consider this the  crude

        beginning of a data base, which it is.

             In  the next few instructions of the program we  assign

        new  values to some of the fields to illustrate  the  method



                                  Page 74









                      Chapter 11 - Structures and Unions


        used to accomplish this.  It should be self explanatory,  so

        no additional comments will be given.

                        A NOTE TO PASCAL PROGRAMMERS

             Pascal  allows  you to copy an entire RECORD  with  one

        statement.   This is not possible in C.   You must copy each

        element  of a structure one at a time.   As improvements  to

        the   language  are  defined,   this  will  be  one  of  the

        refinements.   In fact,  some of the newer compilers already

        allow   structure   assignment.     Check   your    compiler

        documentation to see if your compiler has this feature yet.

                   WE FINALLY DISPLAY ALL OF THE RESULTS

             The last few statements contain a for loop in which all

        of  the generated values are displayed in a formatted  list. 

        Compile  and  run  the program to see if it  does  what  you

        expect it to do.

                   USING POINTERS AND STRUCTURES TOGETHER

             Load  and  display  the  file named  STRUCT3.C  for  an

        example of using pointers with structures.   This program is

        identical  to the last program except that it uses  pointers

        for some of the operations.

             The  first  difference shows up in  the  definition  of

        variables  following  the  structure  definition.   In  this

        program  we define a pointer named "point" which is  defined

        as  a  pointer that points to the structure.   It  would  be

        illegal  to  try to use this pointer to point to  any  other

        variable  type.   There  is a very definite reason for  this

        restriction  in  C as we have alluded to  earlier  and  will

        review in the next few paragraphs.

             The next difference is in the for loop where we use the

        pointer  for accessing the data fields.   Since "kids" is  a

        pointer variable that points to the structure, we can define

        "point"  in  terms  of "kids".   The variable  "kids"  is  a

        constant so it cannot be changed in value,  but "point" is a

        pointer  variable  and can be assigned any value  consistent

        with  its being required to point to the structure.   If  we

        assign  the  value of "kids" to "point" then  it  should  be

        clear  that it will point to the first element of the array,

        a structure containing three fields.

                             POINTER ARITHMETIC

             Adding  1 to "point" will now cause it to point to  the

        second  field of the array because of the way  pointers  are



                                  Page 75









                      Chapter 11 - Structures and Unions


        handled in C.   The system knows that the structure contains

        three  variables  and it knows how many memory elements  are

        required to store the complete structure.   Therefore if  we

        tell it to add one to the pointer,  it will actually add the

        number  of  memory  elements  required to get  to  the  next

        element of the array.   If, for example, we were to add 4 to

        the  pointer,  it would advance the value of the  pointer  4

        times the size of the structure,  resulting in it pointing 4

        elements  farther  along the array.   This is the  reason  a

        pointer  cannot be used to point to any data type other than

        the one for which it was defined.

             Now to return to the program displayed on your monitor. 

        It  should be clear from the previous discussion that as  we

        go through the loop, the pointer will point to the beginning

        of  one of the array elements each time.   We can  therefore

        use  the  pointer to reference the various elements  of  the

        structure.   Referring to the elements of a structure with a

        pointer occurs so often in C that a special method of  doing

        that  was  devised.   Using "point->initial" is the same  as

        using  "(*point).initial" which is really the way we did  it

        in the last two programs.   Remember that *point is the data

        to  which  the pointer points and the  construct  should  be

        clear.   The  "->"  is  made up of the minus  sign  and  the

        greater than sign.

             Since the pointer points to the structure, we must once

        again  define which of the elements we wish to refer to each

        time  we use one of the elements of  the  structure.   There

        are, as we have seen, several different methods of referring

        to  the members of the structure,  and in the for loop  used

        for output at the end of the program, we use three different

        methods.   This  would  be considered very poor  programming

        practice,  but  is done this way here to illustrate  to  you

        that  they all lead to the same result.   This program  will

        probably   require   some  study  on  your  part  to   fully

        understand,  but  it will be worth your time and  effort  to

        grasp these principles.

             Compile and run this program.

                        NESTED AND NAMED STRUCTURES

             Load and display the file named NESTED.C for an example

        of  a nested structure.   The structures we have seen so far

        have been very simple,  although useful.   It is possible to

        define  structures  containing dozens and even  hundreds  or

        thousands  of  elements but it would be to  the  programmers

        advantage not to define all of the elements at one pass  but

        rather to use a hierarchical structure of definition.   This

        will be illustrated with the program on your monitor.



                                  Page 76









                      Chapter 11 - Structures and Unions



             The  first  structure  contains three elements  but  is

        followed by no variable name.  We therefore have not defined

        any variables only a structure, but since we have included a

        name  at the beginning of the structure,  the  structure  is

        named  "person".   The name "person" can be used to refer to

        the  structure  but not to any variable  of  this  structure

        type.   It is therefore a new type that we have defined, and

        we can use the new type in nearly the same way we use "int",

        "char",  or  any  other  types that exist in  C.   The  only

        restriction is that this new name must always be  associated

        with the reserved word "struct".

             The  next  structure definition contains  three  fields

        with the middle field being the previously defined structure

        which we named "person".  The variable which has the type of

        "person" is named "descrip".   So the new structure contains

        two   simple   variables,   "grade"  and  a   string   named

        "lunch[25]",  and  the  structure  named  "descrip".   Since

        "descrip"  contains  three  variables,   the  new  structure

        actually contains 5 variables.  This structure is also given

        a name "alldat",  which is another type definition.  Finally

        we  define an array of 53 variables each with the  structure

        defined by "alldat",  and each with the name "student".   If

        that is clear,  you will see that we have defined a total of

        53 times 5 variables,  each of which is capable of storing a

        value.

                             TWO MORE VARIABLES

             Since  we  have a new type definition we can use it  to

        define  two  more variables.   The variables  "teacher"  and

        "sub"  are defined in the next statement to be variables  of

        the  type  "alldat",  so that each of  these  two  variables

        contain 5 fields which can store data.

                       NOW TO USE SOME OF THE FIELDS

             In  the next five lines of the program,  we will assign

        values to each of the fields of "teacher".   The first field

        is  the  "grade" field and is handled just  like  the  other

        structures  we  have studied because it is not part  of  the

        nested structure.  Next we wish to assign a value to her age

        which  is  part of the nested structure.   To  address  this

        field  we start with the variable name "teacher" to which we

        append  the name of the group "descrip",  and then  we  must

        define which field of the nested structure we are interested

        in,  so  we append the name "age".   The teachers status  is

        handled in exactly the same manner as her age,  but the last

        two  fields  are  assigned  strings using  the  string  copy

        "strcpy" function which must be used for string  assignment. 



                                  Page 77









                      Chapter 11 - Structures and Unions


        Notice  that the variable names in the "strcpy" function are

        still variable names even though they are made up of several

        parts each.

             The variable "sub" is assigned nonsense values in  much

        the  same  way,  but in a different order since they do  not

        have to occur in any required order.   Finally, a few of the

        "student"  variables  are assigned values  for  illustrative

        purposes  and  the program ends.   None of  the  values  are

        printed  for illustration since several were printed in  the

        last examples.

             Compile  and run this program,  but when you run it you

        may get a "stack overflow" error.   C uses it's own internal

        stack  to  store  the  automatic variables  on  but  most  C

        compilers  use only a 2048 byte stack as a  default.    This

        program  has more than that in the defined structures so  it

        will be necessary for you to increase the stack size.    The

        method  for  doing this for some compilers is given  in  the

        accompanying COMPILER.DOC file with this tutorial.   Consult

        your compiler documentation for details about your compiler. 

        There  is  another way around this problem,  and that is  to

        move the structure definitions outside of the program  where

        they will be external variables and therefore static.    The

        result  is that they will not be kept on the internal  stack

        and  the  stack will therefore not overflow.    It would  be

        good for you to try both methods of fixing this problem.

                           MORE ABOUT STRUCTURES

             It is possible to continue nesting structures until you

        get  totally confused.   If you define  them  properly,  the

        computer  will  not get confused because there is no  stated

        limit as to how many levels of nesting are  allowed.   There

        is probably a practical limit of three beyond which you will

        get confused, but the language has no limit.  In addition to

        nesting, you can include as many structures as you desire in

        any level of structures,  such as defining another structure

        prior  to  "alldat" and using it in "alldat" in addition  to

        using  "person".   The  structure named  "person"  could  be

        included in "alldat" two or more times if desired,  as could

        pointers to it.

             Structures can contain arrays of other structures which

        in  turn  can  contain  arrays  of  simple  types  or  other

        structures.   It  can go on and on until you lose all reason

        to  continue.   I am only trying to illustrate to  you  that

        structures  are  very valuable and you will find them  great

        aids to programming if you use them wisely.  Be conservative

        at first, and get bolder as you gain experience.




                                  Page 78









                      Chapter 11 - Structures and Unions


             More  complex structures will not be illustrated  here,

        but  you will find examples of additional structures in  the

        example  programs  included  in the  last  chapter  of  this

        tutorial.     For   example,   see   the   "#include"   file

        "STRUCT.DEF".

                              WHAT ARE UNIONS?

             Load the file named UNION1.C for an example of a union. 

        Simply stated,  a union allows you a way to look at the same

        data  with  different types,  or to use the same  data  with

        different names. Examine the program on your monitor.

             In this example we have two elements to the union,  the

        first part being the integer named "value",  which is stored

        as  a  two byte variable somewhere in the computers  memory. 

        The  second  element is made up of two  character  variables

        named "first" and "second".   These two variables are stored

        in  the  same storage locations that "value" is  stored  in,

        because  that is what a union does.   A union allows you  to

        store  different types of data in the same physical  storage

        locations.  In this case, you could put an integer number in

        "value",  then retrieve it in its two halves by getting each

        half  using  the  two  names  "first"  and  "second".   This

        technique is often used to pack data bytes together when you

        are,  for  example,  combining  bytes  to  be  used  in  the

        registers of the microprocessor. 

             Accessing  the fields of the union are very similar  to

        accessing the fields of a structure and will be left to  you

        to determine by studying the example.

             One  additional  note  must  be given  here  about  the

        program.  When it is run using most compilers, the data will

        be  displayed  with two leading f's due to  the  hexadecimal

        output  promoting  the  char  type  variables  to  int   and

        extending  the  sign bit to the left.   Converting the  char

        type data fields to int type fields prior to display  should

        remove the leading f's from your display.  This will involve

        defining  two new int type variables and assigning the  char

        type  variables to them.   This will be left as an  exercise

        for  you.   Note that the same problem will come up in a few

        of the later files also.

             Compile and run this program and observe that the  data

        is  read out as an "int" and as two "char"  variables.   The

        "char" variables are reversed in order because of the way an

        "int" variable is stored internally in your computer.  Don't

        worry  about this.  It is not a problem but it can be a very

        interesting area of study if you are so inclined.




                                  Page 79









                      Chapter 11 - Structures and Unions


                           ANOTHER UNION EXAMPLE

             Load  and  display the file named UNION2.C for  another

        example of a union,  one which is much more common.  Suppose

        you  wished to build a large database including  information

        on many types of vehicles.  It would be silly to include the

        number of propellers on a car,  or the number of tires on  a

        boat.   In  order to keep all pertinent data,  however,  you

        would  need  those  data points for their  proper  types  of

        vehicles.   In  order to build an efficient data  base,  you

        would need several different types of data for each vehicle,

        some  of which would be common,  and some of which would  be

        different.  That is exactly what we are doing in the example

        program on your monitor.

             In this program,  we will define a complete  structure,

        then  decide which of the various types can go into it.   We

        will  start at the top and work our  way  down.   First,  we

        define  a  few constants with the #defines,  and  begin  the

        program  itself.   We define a structure named  "automobile"

        containing  several fields which you should have no  trouble

        recognizing, but we define no variables at this time.

                         A NEW CONCEPT, THE TYPEDEF

             Next  we  define a new type of data with  a  "typedef". 

        This  defines  a complete new type that can be used  in  the

        same way that "int" or "char" can be used.   Notice that the

        structure  has  no name,  but at the end where  there  would

        normally be a variable name there is the name "BOATDEF".  We

        now have a new type, "BOATDEF", that can be used to define a

        structure anyplace we would like to.   Notice that this does

        not  define  any  variables,  only a  new  type  definition.

        Capitalizing  the name is a personal preference only and  is

        not  a  C standard.   It makes the "typedef" look  different

        from a variable name.

             We  finally come to the big structure that defines  our

        data using the building blocks already defined  above.   The

        structure is composed of 5 parts, two simple variables named

        "vehicle" and "weight",  followed by the union,  and finally

        the last two simple variables named "value" and "owner".  Of

        course  the union is what we need to look at carefully here,

        so focus on it for the moment.   You will notice that it  is

        composed  of four parts,  the first part being the  variable

        "car" which is a structure that we defined previously.   The

        second  part is a variable named "boat" which is a structure

        of the type "BOATDEF" previously defined.  The third part of

        the  union is the variable "airplane" which is  a  structure

        defined in place in the union.   Finally we come to the last





                                  Page 80









                      Chapter 11 - Structures and Unions


        part  of  the  union,  the variable named  "ship"  which  is

        another structure of the type "BOATDEF".

             I  hope  it is obvious to you that all four could  have

        been defined in any of the three ways shown,  but the  three

        different  methods  were used to show you that any could  be

        used.   In practice,  the clearest definition would probably

        have occurred by using the "typedef" for each of the parts.

                            WHAT DO WE HAVE NOW?

             We  now have a structure that can be used to store  any

        of  four different kinds of data structures.   The  size  of

        every  record will be the size of that record containing the

        largest  union.   In this case part 1 is the  largest  union

        because  it is composed of three integers,  the others being

        composed  of  an integer and a character  each.   The  first

        member  of this union would therefore determine the size  of

        all structures of this type.  The resulting structure can be

        used to store any of the four types of data, but it is up to

        the  programmer  to  keep track of what is  stored  in  each

        variable of this type.   The variable "vehicle" was designed

        into  this  structure to keep track of the type  of  vehicle

        stored here.   The four defines at the top of the page  were

        designed  to  be  used  as indicators to be  stored  in  the

        variable "vehicle".
       
             A  few  examples of how to use the resulting  structure

        are given in the next few lines of the program.  Some of the

        variables are defined and a few of them are printed out  for

        illustrative purposes.

             The union is not used too frequently,  and almost never

        by   beginning   programmers.    You   will   encounter   it

        occasionally  so  it is worth your effort to at  least  know

        what  it is.   You do not need to know the details of it  at

        this time,  so don't spend too much time studying it.   When

        you do have a need for a variant structure, a union, you can

        learn it at that time. For your own benefit, however, do not

        slight the structure. You should use the structure often.







        PROGRAMMING EXERCISES

        1.   Define a named structure  containing a string field for

             a name,  an integer for feet, and another for arms. Use



                                  Page 81









                      Chapter 11 - Structures and Unions


             the new type to define an array of about 6 items.  Fill

             the fields with data and print them out as follows.

             A human being has 2 legs and 2 arms.
             A dog has 4 legs and 0 arms.
             A television set has 4 legs and 0 arms.
             A chair has 4 legs and 2 arms.
             etc.

        2.   Rewrite  exercise 1 using  a pointer to print the  data

             out.