1.4.1 - Primitive data types and character sets

Values and their data types

Every stored value needs a meaning before a program can use it.

The same visible symbols can mean different things: adding the integers 12 and 3 gives 15, while concatenating the strings "12" and "3" gives "123". Stored bits need an interpretation; the bits alone do not announce whether they represent a number or text.

The primitive types named in this topic are:

A real number is a mathematical idea; floating point is a finite computer representation for real values. Not every real value can be represented exactly. A Boolean is not the string "False": that string contains five characters, whereas a Boolean false value controls a logical decision. Type names and implementations vary between languages. Python has int, float, str and bool; it uses a length-one string for a character rather than a separate character type.

Choose types from the required use

Choose a type from what the program must do with a value, not from its appearance alone. A school stores a pupil's admission identifier "00418" as a string: the leading zeros matter and the identifier is not added to another identifier. The number of pupils in a class is an integer because pupils are counted in whole units. A mean journey time of 12.6 minutes is real because the calculated result may have a fractional part. A single grade letter can be a character, and a flag recording whether consent was given can be Boolean.

A useful justification connects the type to a requirement. “String because it is text” is less informative than “String because the identifier's leading zeros must be preserved and no arithmetic is required.” Conversely, storing a quantity as text makes numeric comparison or arithmetic require conversion. For example, string order need not agree with numeric order: "100" comes before "20" in a simple character-by-character ordering because 1 comes before 2.

Type choice is separate from validation. An integer can still be an invalid class size, such as -8. A string can still be an invalid identifier. The type gives the category of value; checks such as a permitted range or format decide whether that particular value is acceptable.

Convert input before using it

Casting converts a value to another type when a valid conversion exists. Python's input(prompt) returns a string, even if the user types digits. int converts suitable integer text to an integer, float converts suitable numeric text to a floating-point value, and str produces a string representation. print displays a value; displaying it does not change the stored variable's type.

This is a static example. The user types 7 at the prompt.

text_count = input("Number of tickets: ")
count = int(text_count)
total = count * 2.5
message = "Cost: " + str(total)
print(message)

The states are text_count = "7", count = 7, total = 17.5, and message = "Cost: 17.5". Multiplication uses a number; concatenation uses strings. Without the conversion, "7" + "2" would join text rather than calculate nine.

Conversions can fail or lose information. int("7.5") fails because that text is not an integer literal; float("7.5") succeeds. Converting the already numeric value 7.5 with int(7.5) gives 7, truncating toward zero, not rounding to the nearest integer. Negative int(-7.5) gives -7. An invalid entry such as "seven" needs handling before the program can continue normally. Casting is therefore not a substitute for validating that input has the required format and range.

When checking a short fragment, follow both each value and its type. Convert a numeric count before calculating with it, then convert the calculated amount to text when constructing a message. An identifier can pass through the calculation unchanged as a string, preserving its leading zeros. Python and equivalent pseudocode can express the same conversion steps.