Types

1
Types

• CGS 3460, Lecture 21
• Feb 27, 2006
• Hen-I Yang

2
Announcement

• Quiz 3 will be held on Mar 8th (Wed) instead of
  Mar 10th (Fri).
• Quiz 3 includes homework 3, textbook up to Sec
  7.1, and slides up to last Friday (2/24)
• Always put return 0 as the last statement in your
  code
• TA will post a couple sample codes to help you
  learn nested loop.
• It is strongly discouraged to simply use hundreds
  of statements for Homework 3.

3
Previously

• Quiz 2
• int

4
Agenda

• int
• float
• char
• sizeof
• Type conversion

5
float

• float, double, long double
• 3 parts sign, exponent and fraction
• float 1.17 x 10-38 3.4 x 1038
• 6 digits precision
• double 2.22 x 10-308 1.79 x 10308
• 15 digits precision
• long double 3.36 x 10-4931 - 1.19 x 104932
• Its all about number of bits.
• float (single precision) uses 32 bits
• Usually long double ranges from 80 to 132 bits
• By default, the values are stored as double
  precision number

6
float (II)

• The range of values are machine dependent
• When specified as a constant expression
• 57.0F for float
• 57.0L for long double
• More conversion specifications
• lf for double
• Lf for long double

7
char

• ASCII (7 bits 128 characters)
• char a 0
• C treats characters as small integers
• It is possible to use char in a lot of places
  where you use int
• signed/unsigned

8
Char (II)
9
Escape sequence

• The escape sequences we knew
• Octal escape sequence
• Hexadecimal escape sequence

Reading/Writing characters

• Conversion specification c
• getchar(), putchar(ch)
• They are faster

10
sizeof()

• Tell you how many bytes a variable occupies

11
Type Conversion

• Allows mixing of basic types.
• Compiler takes care of the dirty work. (implicit
  conversions)
• Coercion
• happens when (usual arithmetic conversions)
• operands are of different types
• right hand side does not match result of the left
  hand side
• argument declaration and invocation
• return type

12
Problem with Type Conversion

• int ? float
• may lose precision
• float ? int
• not a good idea
• long ? int
• overflow problem

13
Type Conversion (II)

• Strategy convert operands to the narrowest
  type that can safely accommodate both values, aka
  promotion.
• Promotion if either operand is a floating type
• float ? double ? long double
• Promotion if both operands are of integer type
  int ? unsigned int ? long ? unsigned long
• Try not to mixed singed and unsigned integer
• Assignment simply follow the type on the left
  hand side (hence) maybe very troublesome.
• Check out example on page 127.

14
Casting (Explicit Conversion)

• (type-name) expression
• Forced type conversion
• Useful, but also dangerous
• Use of casting to get the value you want
• Use of casting to avoid overflow
• Casting allows programmers to avoid the
  restriction setup by type system, but also expose
  to all sort of vulnerability because of the
  removal of such restrictions

15
typedef

• Typedef int Bool
• Define new type
• After typedef, new type can be used just like
  basic type
• Can define a type with a more proper name (more
  informative)
• Enhance portability
• More powerful than define macro,
• Change once, effective all
• can be used anywhere as synonym of original type
• Obey scope rule

16
Summary

• int
• float
• char
• sizeof
• Type conversion
• typedef

17
Before you go

• Read Chapter 7.
• Exercise 7.15, 7.17
• Homework 3 due Mar 7, Tuesday at 1159 pm
• Quiz 3 will be held on Mar 8th (Wed).
• Quiz 3 covers
• homework 3,
• textbook up to Sec 7.1, and
• slides up to last Friday (2/24)