Use of integers in Chapter 7

Nat · December 3rd, 2009, 08:27 AM

First of all - loving the book. OOP is completely new to me although I have coded before in FORTRAN and a bit of HTML and ASP and I have found the structure and techniques brilliant.

In Chapter 7 on page 168 the code:

Code:

 last = (int) randomNumber.Next(max);

is used. Why is the "(int)" necessary? and why is it not used when assigning to "current" three lines later?

Secondly in the factorial exercise at the end of the chapter could one use a "long" type integer as well as a "double"?

Thanks in advance for answers!

Will · December 3rd, 2009, 05:15 PM

Not got this far so I could be wrong but it is an explicit conversion which casts the value held in last into an int datatype.

Is what I would have said before I looked at the code provided in the book as last is already of type int?

Furthermore if you do happen to cast a variable that is already the same type and you make a mistake will the code, compile or provide a warning?

Will · December 3rd, 2009, 05:23 PM

I just tested this inthe code below:

Code:

 private void button1_Click(object sender, EventArgs e)
        {
            int last;

            Random number = new Random();

            last = number.Next();

            textBox1.Text = last.ToString();

runs exactly the same as:

Code:

 private void button1_Click(object sender, EventArgs e)
        {
            int last;

            Random number = new Random();

            last = (int) number.Next();

            textBox1.Text = last.ToString();

I don't know if I have missed something here but have you tried running the code without the (int) ? does it run as intended?

Bob Bedell · December 3rd, 2009, 10:44 PM

The cast isn't necessary, and the compiler will ignore it. Random.Next() returns an Int32. No cast required.

Both of the blocks of code Will posted compile to 1 line of code:

Code:

int last = new Random().Next();

DrPurdum · December 4th, 2009, 01:22 AM

You're all right...the cast is NOT necessary. However, it reflects a habit I got into when I was programming in C and earlier versions of C#. The random number generator used to return a value between 0 and 1.0 and was a double data type. Since a double requires 8 bytes and an int only uses 4 bytes, this would be an example of data narrowing (see page 67). Trying to assign an 8 byte thingie into a 4 byte thingie (i.e., data narrowing) runs the risk of losing information, you should always use a cast. It's like trying to pour 8 gallons of data into a 4 gallon bucket; some of the data is going to slop over the sides of the smaller bucket. That's why casts must always be used when data narrowing.

If you reverse the assignment direction; that is, assigning a 4 byte object into an 8 byte object, this is called data widening. Pouring 4 gallons of data into an 8 gallon bucket runs no risk of data lost and the use of a cast is not required. However, I always try to code so that I use a cast when data objects do not match. In other words, the cast is also a form of documentation that tells the reader of my code that I really intended to do this. In commercial code, I even do this with data widening. (There's no performance hit, since the compiler throws it away.)

The cast in this example is not necessary at all since the new random number generator defaults to returning an int data type. My mind was probably stuck with the old versions I used to use that always returned a double. Always keep in mind, however, that casts should always be used when assigning different data types, if for no other reason that it documents what you're doing. I think it's a good idea to always use casts whenever the data types don't match.

DrPurdum · December 4th, 2009, 01:29 AM

I forgot to answer the factorial question. Because factorial are normally integral values, you could a long for the data type. However, even as big as a long is, relatively small numbers overflow the factorial calculation quickly. (If I remember correctly, the number is less than 100! to overflow a long.) If you need really big factorials, you can use Sterlings Approximation with a double data type.