Variables

Variables are convenient handles for accessing memory. We don't need to mess with memory addresses:

Figure 66. Variables: Handles to memory

Declaring a variable requires a type name like double and an identifier:

// Variable declaration:
//     Variable's type is double
//     Variable's name is «pi» (identifier)

double pi;

{type name} {variable name} ;

We may assign values to variables or build expressions like pi * 2.0 * 2.0 :

Figure 68. Declare, assign and use

double pi;      // Variable declaration

pi = 3.1415926; // Assigning value to variable

// Print a circle's area of radius 2.0

System.out.println(pi * 2.0 * 2.0);

Figure 69. Combining declaration and initialization

Separate declaration and initialization

double pi;      // Declaration of variable pi

pi = 3.1415926; // Value assignment

Combined declaration and initialization

double pi = 3.1415926;

Figure 70. Multiple variables of same type

int a;
int b = 22;
int c;

being equivalent to either of:

Compact	Multiple lines
`int a, b = 22, c;`	`int a, b = 22, c;`

Figure 71. Identifier in Java™:

Variable names.
Class names
Method names, e.g.:

public static void main(String [] args)

Figure 72. Identifier name examples:

Identifier rules	Legal	Illegal
Start with a letter, “`_`” or “`$`” May be followed by letters, digits, “`_`” or “`$`” Must not match a reserved keyword or literal	`$test` `_$` `count` `blue`	`2sad` `switch` `true`

Figure 73. Java™ keywords.

Reserved keywords

abstract  char      else     if          long       return        this      while
assert    class     enum     goto        native     short         throw     _ (underscore)
boolean   const     extends  implements  new        static        throws  
break     continue  final    import      package    strictfp       transient  
byte      default   finally  instanceof  private    super         try  
case      do        float    int         protected  switch        void  
catch     double    for      interface   public     synchronized  volatile

Reserved boolean and null literal values

true   false   null

Contextual keywords

exports  non-sealed opens    provides  requires  to          uses  when  yield
module   open       permits  record    sealed    transitive  var   with

Figure 74. Variable naming conventions

Start with a small letter like africa rather than Africa.

Use “camel case” e.g. myFirstCode.

Do not start with _ or $.

Figure 75. Constant variables

Modifier final prohibits changes:

final double PI = 3.1415926;
...
PI = 1.0; // Compile time error: Constant cannot be modified

Note

final variables by convention are being capitalized

Figure 76. Case sensitivity

Variable names are case sensitive:

int count = 32;
int Count = 44;
System.out.println(count + ":" + Count);

Resulting output:

32:44

exercise No. 11

Legal variable names

Which of the following names are legal variable names? If so, would you actually use them in your code?

Complete the following table and explain your decision with respect to the “Language Fundamentals” / “variables” section of [Kurniawan] .

Identifier	Legal? (yes / no)	Explanation / remark
`for`
`sum_of_data`
`sumOfData`
`first-name`
`ABC`
`42isThesolution`
`println`
`B4`
`AnnualSalary`
`"hello"`
`_average`
`ανδρος`
`$sum`

Tip

You may want to prepare a simple Java™ program testing the above names.

Consider:

public static void main(String[] args) {
   int 42isThesolution = 6; // Syntax error on token "42", delete this token
}

Unfortunately this message does not explain the underlying cause: It does not hint towards an illegal identifier name.

One indication of a sound compiler implementation is its ability to provide meaningful, self explanatory error messages.

Identifier	Legal? (yes / no)	Explanation / remark
`for`	no	“for” is a Java keyword.
`sum_of_data`	yes	-
`sumOfData`	yes	-
`first-name`	no	Operators like “-” or “+” must not appear in variable names.
`ABC`	yes	Discouraged by «Best practices» if non-constant (`final`): Non-constant variables should start with lowercase letters.
`42isThesolution`	no	Identifiers must not start with a number.
`println`	yes	`println` is a method's name but is no Java™ keyword.
`B4`	yes	See «ABC» above.
`AnnualSalary`	yes	See «ABC» above.
`"hello"`	no	String delimiters must not be part of an identifier.
`_average`	yes	Discouraged by «Best practices»: Identifier starting with «_» should be reserved for system code.
`ανδρος`	yes	Perfectly legal Greek Unicode characters.
`$sum`	yes	Discouraged by «Best practices»: Identifier starting with «$» should be reserved for system code.

Figure 77. Define before use

Correct:

double f;
f = -4.55;

Wrong:

f = -4.55;
double f;

Figure 78. Type safety

int i = 2;
int j = i;   // o.K.: Assigning int to int
long l = i;  // o.K.: Widening conversion

i = l;       // Wrong: Narrowing

boolean b = true;
i = b;       // Error: int and boolean are incompatible types
i = 4.3345   // Error: Cannot assign double to int
i = "Hello"; // Even worse: Assigning a String to an int

Figure 79. Compile time analysis

`byte b127 = 127; // o.K., static check byte b128 = 128; // Wrong: Exceeding 127`	Performing static range check
`int a = 120; byte b120 = a; // Error Incompatible types // Required: byte // Found:int`	No static check on `int` variable's value

exercise No. 12

Benefits of `final`

We reconsider our code from Figure 79, “Compile time analysis ”:

int a = 120;

byte b120 = a; // Error Incompatible types
               // Required: byte
               // Found:int

Adding final solves the issue:

final int a = 120;

byte b120 = a;  // o.K.

Why does introducing a final modifier solves the fatal compile time error issue?

Tip

Read Static program analysis. Why does final help the compiler in code analysis?

Modifying variable a to become final assures its initial value will never change. Thus from a Data-flow analysis point of view the two snippets are equivalent:

final int a = 120;

byte b120 = a;

byte b120 = 120;

The compiler is thus able to perform a compile time range check assuring the value of 120 to be within the interval [-128, 127].

Figure 80. Type inference (JDK™ 10)

var count = 30; // o.K., type can be inferred from int literal
        
var index;      // Wrong: No way to infer type here.
index = 3;

Figure 81. Forcing conversions

long l = 4345;

int i = (int) l; // Casting long to int
System.out.println("i carrying long: " + i);

double d = 44.2323;
i = (int) d; // Casting double to int
System.out.println("i carrying double: " + i);

i carrying long: 4345
i carrying double: 44

Figure 82. Watch out!

long l = 3000000000000000L;
int i = (int) l;
System.out.println("i carrying long:" + i);

double d = 44300000000000.0;
i = (int) d;
System.out.println("i carrying double:" + i);

i carrying long:-296517632
i carrying double:2147483647

Figure 83. Casting long to int

‹›

Figure 84. Don't worry, be happy ...

«C» programming language miracles:

#include <stdio.h>

void main(void) {
  double measure = 65234.5435;
  short velocity;
  velocity = measure;
  printf("Velocity=%d\n", velocity);
}

Uups:

Velocity=-302

Figure 85. ... and watch the outcome

Development costs:

~$7 billion.

Rocket and cargo

~$500 million.

«C» vs. Java.

We reconsider the «C» code example from Figure 84, “Don't worry, be happy ... ”:

#include <stdio.h>

void main(void) {
  double measure = 65234.5435;
  short velocity;
  velocity = measure;
  printf("Velocity=%d\n", velocity);
}

Rewrite the same code in Java. Replace printf("Velocity=%d\n", velocity) by a System.out.println(...) statement. Answer the following two questions:

You will encounter a compile time error not being present in »C«. Try to overcome it.

Tip

Read Object Type Casting in Java.

After your correction the result will be -302 like in the »C« snippet. On contrary for a related double to int conversion the overflow behaviour is different:

Code	Output
double d = 3000000000.0; // Larger than Integer.MAX_VALUE int i = (int) d; System.out.println(i); // Result: Integer.MAX_VALUE	2147483647

This time we get an int's maximum value representing the closest possible value to 3000000000.0.

If the velocity = measure assignment was behaving accordingly we should see an outcome of Short.MAX_VALUE or 32767 rather than -302.

Explain this different behaviour.

Tip

Read 5.1.3. Narrowing Primitive Conversion.

Naively rewriting the »C« snippet in Java™ yields a compile time error:
```
double measure = 65234.5435;
short velocity;
velocity = measure; // Error: Incompatible types. Required: short, Found: double
System.out.println(velocity);
```
Resolving this error requites a type cast. We also combine declaration and assignment of velocity:
```
double measure = 65234.5435;
short velocity = (short) measure; // Explicit double to short conversion, possible data loss
```
Note

This resolves the syntax error. However the overflow problem when converting a double 65234.5435 into a short of -302 still prevails.

The referenced specification reads:

A narrowing conversion of a floating-point number to an integral type T takes two steps:

In the first step, the floating-point number is converted either to a long, if T is long, or to an int, if T is byte, short, char, or int as follows:

...
In the second step:
- If T is int or long, the result of the conversion is the result of the first step.
  
  This is the behaviour for double to int.
- If T is byte, char, or short, the result of the conversion is the result of a narrowing conversion to type T (§5.1.3) of the result of the first step.
  
  This is the behaviour for double to short.

So behind the scenes the velocity = (short) measure assignment is being decomposed into two steps:

Code

Output

double measure = 65234.5435;

// Step 1: double --> int
int velocityInt = (int) measure;

// Step 2: int --> short
short velocity = (short) velocityInt;

System.out.println("velocityInt=" + velocityInt + ", velocity=" + velocity);

velocityInt=65234, velocity=-302

exercise No. 14

Assignment and type safety

We consider:

int i = -21;
double d = i;

This snippet shows correct Java™ code. However Figure 78, “Type safety ” suggests we should not be allowed to assign an int to a double value due to type safety.

Even worse: Swapping types yields a compile time error:

double d = 2.4;
int i = d; // Error: Incompatible types

Questions to answer:

Explain on syntax level why is d = i is allowed but i = d causes an error? Hint: Read about widening and narrowing conventions.
Whats the ratio behind this design decision? Hint: What would happen for larger double values?

The first code snippet uses the widening conversion: When assigning d = i the Java™ compiler implicitly converts a four byte int into an eight byte double value.

In contrast there is no automated narrowing from double to int hence causing a compile time error.
Turning a double into an int is more cumbersome: The expression e.g. i = 3.5 could be evaluated by agreeing on a specific rounding prescription. But what about e.g. i = 3457357385783573478955345.45 ? A Java™ int's maximum value is $2^{31} - 1$ unable to accommodate the double value in question by far.

Conclusion: Java™ disallows double to int assignments unless using a so called cast (explicit type conversion):

Caveats using explicit casts:

double d = 2.6;
int i = (int) d; // Explicit cast double to int
System.out.println("Result: " + i);

Result: 2

Notice the result of 2 rather than 3 as you might have expected assuming common rounding convention rules. Rounding does not happen unless you explicitly use methods like e.g. Math.round().

double d = 3_000_000_000.;
int i = (int) d; // Explicit cast double to int
System.out.println("Result: " + i);

Result: 2147483647

This truncation of 3 billion to 2147483647 is due to the latter value being the largest possible int value to be represented in Java™.

Java™ provides meta information on types:

exercise No. 15

Inventing `tinyint`.

Suppose Java™ was offering a hypothetic signed integer type tinyint using 3-bit 2-complement representation.

Write down all possible decimal values among with their corresponding binary representation.
Perform the following additions on binary level:
- 1 + 2
- -4 + 2
- -1 - 2
- 3 + 1
Hint by example: -4 + 3 is:
```
 1   100
 2  +011
--  ----
 ?   111  
```
What is 3-bit two complement 111 in decimal?

Given n=3 in three bit two complement notation as usual possible values range from $- 2^{(3 - 1)}$ to $2^{(3 - 1)} - 1$ :

Decimal	Two-complement
3	`011`
2	`010`
1	`001`
0	`000`
-1	`111`
-2	`110`
-3	`101`
-4	`100`

Thus the example hint value 111 is indeed decimal -1 equalling -4 + 3 as expected.

1 + 2
```
 1   001
 2  +010
--  ----
 3   011
```
-4 + 2
```
-4   100
 2  +010
--  ----
-2   110
```
-1 - 2
```
-1     111
-2    +110
--  ------
-3    1101 (See comment below)
```
In 3-bit two complement representation the leftmost bit at position 4 will be discarded. The actual result is thus 101 as expected.
3 + 1
```
 3   011
 1  +001
--  ----
-4   100 Surprise! See comment below
```
This is actually an overflow: In 3-bit two complement representation the highest representable decimal value is 3 being equal to $2^{3 - 1} - 1$ . Trying to add a value of 1 leads to disaster and in suitable circumstances may cause a US$370 spacecraft plummeting from the sky! Excerpt from the Ariane 5 flight 501 failure report:

The internal SRI software exception was caused during execution of a data conversion from 64-bit floating point to 16-bit signed integer value. The floating point number which was converted had a value greater than what could be represented by a 16-bit signed integer.

exercise No. 16

An `int`'s minimum and maximum value

In this exercise we look at an int's the largest and smallest (negative) possible value.

A Java™ int is internally being represented by 4 bytes. ${00000000_00000000_00000000_00000101}_{2}$ for example represents the decimal value 5.

Java™ implements negative values using Two's complement representation. We provide some values:

Table 1. 4 Byte Two's complement representation of int values.

Two complement representation	Decimal representation
`00000000_00000000_00000000_00000000`	0
`01111111_11111111_11111111_11111111`	$2^{32 - 1} - 1$ (Maximum)
`10000000_00000000_00000000_00000000`	$- 2^{32 - 1}$ (Minimum)
`11111111_11111111_11111111_11111111`	-1

Use int literals in binary representation like e.g. 0B1100 in the “Language Fundamentals” / “Literals” section of [Kurniawan] in order to write an int's minimum and maximum possible value to standard output.

public static void main(String[] args) {

  int minumum = ... , //TODO: provide values by
  maximum = ...;  // binary int literals

  System.out.println("Minimum:" + minumum);
  System.out.println("Maximum:" + maximum);
}

We insert Two's complement representations of minimum and maximum int values according to Table 1, “4 Byte Two's complement representation of int values.”.

public static void main(String[] args) {

  int minumum = 0B10000000_00000000_00000000_00000000,
      maximum = 0B01111111_11111111_11111111_11111111;

  System.out.println("Minimum int value:" + minumum);
  System.out.println("Maximum int value:" + maximum);
}

BTW: The JDK™ does provide maximum value, minimum value and related information for char, byte, short and int primitive data types within their corresponding Character, Byte, Short and Integer classes. You may want to execute:

System.out.println("int minimum:" + Integer.MIN_VALUE);
System.out.println("int maximum:" + Integer.MAX_VALUE);

System.out.println("int bytes:" + Integer.BYTES);
System.out.println("int size:" + Integer.SIZE);

Figure 87. Dynamic typing in PERL

$test = 44;  # Assigning an int
print $test, "\n";

$test = "Jim"; # Assigning a string
print $test, "\n";

$cmp = 43.55; # A float

if ($test == $cmp) { # comparing string against float
   print "Equal\n";
} else {
   print "Different\n";
}

44
Jim
Different

Figure 88. Dynamic typing in PERL, part 2

$a = 2; # An integer
$b = 3; # Another integer

print '$a + $b = ', $a + $b, "\n";

$jim = "Jim";         # A string
print '$jim + $a = ' , $jim + $a, "\n";

$a + $b = 5
$jim + $a = 2

Figure 89. Using final

//Bad!
double pi = 3.141592653589793;
...
pi = -4; // Woops, accidential and erroneous redefinition

//Good
final double PI = 3.141592653589793;
...
PI = -4; // Compile time error:
         // Cannot assign a value to final variable 'pi'

Figure 90. Reference type examples

GpsPosition start = new GpsPosition(48.7758, 9.1829);
String name = "Simon";
LocalDate birtday = LocalDate.of(1990, Month.JULY, 5);

Prev	Up	Next
Primitive types	Home	Literals

Language Fundamentals

Variables

Note

Legal variable names

Tip

Benefits of final

Tip

«C» vs. Java.

Tip

Tip

Note

Assignment and type safety

Inventing tinyint.

An int's minimum and maximum value

Benefits of `final`

Inventing `tinyint`.

An `int`'s minimum and maximum value