COMP2521 Style Guide

Contents

Recommended Code Layout

Code Layout

Use the following order for laying out your code:

  1. Header comment
  2. Inclusion of system header files
  3. Inclusion of local header files
  4. Definition of constants (#defines and enums)
  5. Definition of data types (structs and typedefs)
  6. Global variables (try to avoid these altogether)
  7. Function prototypes
  8. main() function
  9. Function definitions
Code Layout

Use the following order for laying out your code:

  1. Header comment
  2. Inclusion of system header files
  3. Inclusion of local header files
  4. Definition of constants (#defines and enums)
  5. Definition of data types (structs and typedefs)
  6. Global variables (try to avoid these altogether)
  7. Function prototypes
  8. main() function
  9. Function definitions
// COMP2521 Lab 10 Exercise - Turing's test
//
// Pretend to be a human in text-based conversation.
// https://en.wikipedia.org/wiki/Turing_test
//
// Authors:
// Grace Hopper (z1234567@unsw.edu.au)
// Ada Lovelace (z5000000@unsw.edu.au)
//
// Written: 19/03/2019

#include <stdio.h>
#include <string.h>
#include <math.h>

#include "language.h"

#define MEANING_OF_LIFE 42

// struct definitions would go here

// function prototypes would go here


int main(void) {
    printf("Hello!\n");
    return 0;
}

// function definitions would go here

Header Comment

All programs should have a header comment.

For small lab exercises the header comment should contain at least:

  • The names of the author(s) - include your UNSW zID with any code you submit at UNSW
  • The date it was written
  • A description of what the program does

For larger programs such as assignments, also include:

  • An overview of how the program works, including any bugs or limitations
  • Any other information that might help a reader understand the code
  • References to resources used in constructing the code

It is not necessary to include licensing information, but it is commonly present in header comments of code outside uni.

Header Comment

All programs should have a header comment.

For small lab exercises the header comment should contain at least:

  • The names of the author(s) - include your UNSW zID with any code you submit at UNSW
  • The date it was written
  • A description of what the program does

For larger programs such as assignments, also include:

  • An overview of how the program works, including any bugs or limitations
  • Any other information that might help a reader understand the code
  • References to resources used in constructing the code

It is not necessary to include licensing information, but it is commonly present in header comments of code outside uni.

// COMP2521 Lab 10 Exercise - Turing's test
//
// Pretend to be a human in text-based conversation.
// https://en.wikipedia.org/wiki/Turing_test
//
// Authors:
// Grace Hopper (z1234567@unsw.edu.au)
// Ada Lovelace (z5000000@unsw.edu.au)
//
// Written: 19/03/2019
Braces

Braces must be used with if, else, while, for and do ... while statements, even when the body is empty or contains only a single statement.

Compact braces should be used, which means:

  • No line break before the opening brace.
  • Line break after the opening brace.
  • Line break before the closing brace.
  • Line break after the closing brace, except when it is followed by else or when it is followed by the while in a do ... while statement.
  • There is an exception for the previous point - for long conditions (of if, while, for and do ... while statements) and functions that have long parameter lists (which span multiple lines), the opening brace may appear on its own line.
Braces

Braces must be used with if, else, while, for and do ... while statements, even when the body is empty or contains only a single statement.

Compact braces should be used, which means:

  • No line break before the opening brace.
  • Line break after the opening brace.
  • Line break before the closing brace.
  • Line break after the closing brace, except when it is followed by else or when it is followed by the while in a do ... while statement.
  • There is an exception for the previous point - for long conditions (of if, while, for and do ... while statements) and functions that have long parameter lists (which span multiple lines), the opening brace may appear on its own line.

For functions, the opening brace should be on the same line as the function signature. (unless the function has a long parameter list which spans multiple lines, in which case the opening brace may appear on its own line.) The closing brace should be on its own line and not indented.

For if statements, the opening brace should be on the same line as the if and the condition. The closing brace should be on the same line as the following else, if there is one, or on its own line otherwise, and should be indented to the same level as the matching if.

For while and for loops, the opening brace should be on the same line as the loop condition (unless the condition is long and spans multiple lines, in which case the opening brace may appear on its own line). The closing brace should be on its own line and should be indented to the same level as the matching while/for.

For do ... while loops, the opening brace should be on the same line as the do. The closing brace should on the same line as the while, and should be indented to the same level as the do.

Braces

Braces must be used with if, else, while, for and do ... while statements, even when the body is empty or contains only a single statement.

Compact braces should be used, which means:

  • No line break before the opening brace.
  • Line break after the opening brace.
  • Line break before the closing brace.
  • Line break after the closing brace, except when it is followed by else or when it is followed by the while in a do ... while statement.
  • There is an exception for the previous point - for long conditions (of if, while, for and do ... while statements) and functions that have long parameter lists (which span multiple lines), the opening brace may appear on its own line.
int someFunction(int a, int b) {
    if (a > b) {
        // do something
    } else if (a < b) {
        // do something
    } else {
        // do something
    }

    int i = 0;
    while (i <= 10) {
        // do something
        i++;
    }
    
    for (int j = 10; j >= 0; j--) {
    	// do something
    }
    
    int k = 0;
    do {
        // do something
        k++;
    } while (k <= 10);
}

void functionWithManyParameters(int a, int b, int c, int d, int e, int f,
                                int g, int h)
{
    // do something
}
Braces

Braces must be used with if, else, while, for and do ... while statements, even when the body is empty or contains only a single statement.

Compact braces should be used, which means:

  • No line break before the opening brace.
  • Line break after the opening brace.
  • Line break before the closing brace.
  • Line break after the closing brace, except when it is followed by else or when it is followed by the while in a do ... while statement.
  • There is an exception for the previous point - for long conditions (of if, while, for and do ... while statements) and functions that have long parameter lists (which span multiple lines), the opening brace may appear on its own line.
// BAD - the code between the {} should be indented

if (i % 2 == 0) {
printf("even");
} else {
printf("odd");
}

// BAD - put newlines after { and before and after }

while (i < 0) { i++; }

// BAD - always use braces on if statements and while loops

while (i < 0)
    i++;

// BAD - closing braces don't line up

if (a > b) {
    // do something
    } else if (b < a) {
        // do something
        } else {
            // do something
            }
Indentation

Between any pair of braces, the indentation level should increase by one stop. You may use either tabs or spaces for indentation, but do not use both approaches in the same program.

For long expressions that span multiple lines, the indentation level of each line (except the first) should increase by two stops or by an amount that improves readability (e.g., aligning the arguments of long function calls).

Indentation

Between any pair of braces, the indentation level should increase by one stop. You may use either tabs or spaces for indentation, but do not use both approaches in the same program.

For long expressions that span multiple lines, the indentation level of each line (except the first) should increase by two stops or by an amount that improves readability (e.g., aligning the arguments of long function calls).

if (something && somethingElse && somethingElseAgain && anotherThing &&
        evenMoreThings && somethingElseEntirely && thereAreTooManyConditions &&
        thisLineIsTooLong)
{
    ...
}

functionWithManyArguments(something, somethingElse, somethingElseAgain,
                          anotherThing, evenMoreThings);
Spacing

Include a space after keywords such as:

if, else, while, for, do, return

Include a space on each side of binary operators such as:

=  +  -  <  >  *  /  %   <=  >=  ==  !=

Do not include a space after prefix unary operators such as:

& * !

Do not include a space before postfix unary operators such as:

++ --

Do not include a space on each side of member access operators such as:

. ->
Spacing

Include a space after keywords such as:

if, else, while, for, do, return

Include a space on each side of binary operators such as:

=  +  -  <  >  *  /  %   <=  >=  ==  !=

Do not include a space after prefix unary operators such as:

& * !

Do not include a space before postfix unary operators such as:

++ --

Do not include a space on each side of member access operators such as:

. ->
// This is correct:
if (a + b > c) {
    i++;
    curr = curr->next;
}


// This is *not* correct:
if(a+b>c) {
    i ++;
    curr = curr -> next;
}
Statements

Only one executable statement should be used per line of code.

Statements

Only one executable statement should be used per line of code.

Don't put multiple statements on a single line - it makes code very hard to read.

Line Width

Keep lines under 80 characters.

Line Width

Keep lines under 80 characters.

Break long lines up to keep them under 80 characters, unless keeping it as a longer line is significantly more readable.

Line Width

Keep lines under 80 characters.

// This is WAY too long:
if (something && somethingElse && somethingElseAgain && anotherThing && evenMoreThings && somethingElseEntirely && thereAreTooManyConditions && thisLineIsTooLong) {

// This is better (but consider using a function instead):
if (something && somethingElse && somethingElseAgain && anotherThing &&
        evenMoreThings && somethingElseEntirely && thereAreTooManyConditions &&
        thisLineIsTooLong) {

// This is best:
if (thoseThingsAreTrue()) {

Constants

Constants

Use #defines to give constants names.

enums may be used to give groups of related constants names.

#define and enum names must be written in ALL_CAPS_WITH_UNDERSCORES.

Constants

Use #defines to give constants names.

enums may be used to give groups of related constants names.

#define and enum names must be written in ALL_CAPS_WITH_UNDERSCORES.

Unexplained numbers, often called magic numbers, can be hard to understand.

If a number appears multiple times in the code, bugs are created when the is code changed but not all occurrences of the number are changed.

A similar problem is that a number may appear in the code for multiple reasons, e.g., a commonly used number like 10, and if the code needs to be changed it can be hard to determine which occurrences of the number should be changed.

enum is also used by programmers to give constants names in C.

Constants

Use #defines to give constants names.

enums may be used to give groups of related constants names.

#define and enum names must be written in ALL_CAPS_WITH_UNDERSCORES.

#define DAYS_OF_WEEK 7

typedef enum {
	MONDAY,
	TUESDAY,
	WEDNESDAY,
	THURSDAY,
	FRIDAY,
	SATURDAY,
	SUNDAY,
} Weekday;

// ....

int array[DAYS_OF_WEEK];
int i = 0;
while (i < DAYS_OF_WEEK) {
    a[i] = i;
    i++;
}
Constants

Use #defines to give constants names.

enums may be used to give groups of related constants names.

#define and enum names must be written in ALL_CAPS_WITH_UNDERSCORES.

// BAD - the constant 7 should be given a name
int array[7];
int i = 0;
while (i < 7) {
    a[i] = i;
    i++;
}

Variables

Variable Names

Descriptive variable names should always be used where possible.

Short variable names such as x or i are acceptable if there is no appropriate long descriptive name (which is often the case for variables used as loop counters), the meaning of the variable is obvious, or if the variable is used in a very small scope. Examples of acceptable short variable names are:

  • l for list
  • t for tree
  • g for graph
  • n for node
  • v for vertex or value
  • v and w for a pair of vertices
  • i, j and k for loop counters

Variable names must begin with a lower case letter.

Multi-word variable names be in either snake_case or camelCase. Use only one of these approaches in your program.

Declaring Variables

Variables should be declared when (or close to when) they are first assigned a value. Variables should be declared close to when they are first used.

Declaring Variables

Variables should be declared when (or close to when) they are first assigned a value. Variables should be declared close to when they are first used.

int i = 0;

int j;
if (scanf("%d", &j) == 1) {
    // do something
}
Declaring Variables

Variables should be declared when (or close to when) they are first assigned a value. Variables should be declared close to when they are first used.

int i;

// a lot of other code not involving i

i = 0;
Numeric Variables

Use only the types int and double for numeric variables.

You should not need to use other types such as short and float in this course.

You will need to use the type char for arrays, but do not use the type char for simple variables.

Numeric Variables

Use only the types int and double for numeric variables.

You should not need to use other types such as short and float in this course.

You will need to use the type char for arrays, but do not use the type char for simple variables.

int age = 18;
double height = 1.65;
int c = 'A';
Numeric Variables

Use only the types int and double for numeric variables.

You should not need to use other types such as short and float in this course.

You will need to use the type char for arrays, but do not use the type char for simple variables.

short age = 18;
float height = 1.65;
char c = 'A';

// common C programming bug - getchar returns an int
char c = getchar();
Loop Variables

It is recommended variables controlling while loops be declared before the loop and used only to control that loop.

It is recommended variables controlling for loops be declared as part of the for loop.

Loop Variables

It is recommended variables controlling while loops be declared before the loop and used only to control that loop.

It is recommended variables controlling for loops be declared as part of the for loop.

int i = 0;
while (i < MAXIMUM) {
    // do something
    i++;
}

for (int i = 0; i < MAXIMUM; i++) {
    // do something
}
Global and Static Variables

Global variables and static variables should not be used unless you are explicitly permitted.

Global and Static Variables

Global variables and static variables should not be used unless you are explicitly permitted.

Global variables are variables that are declared outside of a function.

Global variables make code hard to read, maintain and parallelize.

There are important uses for global variables but not in this course.

The only acceptable use likely in this course would be a global variable to control debug output.

All the above also applies to variables inside functions which have the static qualifier.

The values in static variables persist across multiple calls to the function creating similar problems.

There is likely to be a major style deduction if you use global variables or static variables without permission.

Variable-Length Arrays

Variable-length arrays should not be used in this course.

Variable-Length Arrays

Variable-length arrays should not be used in this course.

Variable-length arrays are arrays whose size is not known when the program compiles.

Arrays should only be declared with a constant size, such as with a literal integer, e.g. int array[12], or with a constant, e.g. int array[MAX_LENGTH]. Arrays should not be declared with a size using a variable, such as int array[x].

Many C implementations poorly handle local variables consuming more than a small amount of memory, often terminating with a cryptic message if the storage used for local variables exceeds a few megabytes.

This applies to both arrays with a fixed size at compile-time and variable-length arrays.

For this reason, programmers often use malloc for larger arrays (10000+ elements). malloc gives the programmer complete control over storage lifetime, but as a consequence introduces other classes of bugs.

Comments

Comment Use

Comments should be used to describe, explain and/or justify code; they should not simply restate what the code is doing.

Comment Use

Comments should be used to describe, explain and/or justify code; they should not simply restate what the code is doing.

// If the tree is empty
if (tree == NULL) {
    ...
}
Comment Use

Comments should be used to describe, explain and/or justify code; they should not simply restate what the code is doing.

// If the given pointer points to NULL
if (tree == NULL) {
    ...
}

// Declare a file pointer
FILE *file;

// Initialise sum as 0
int sum = 0;

// Initialise wordList as NULL
List wordList = NULL;

// Increment count
count++;

// Declare word name variable of max 100 characters
char word[MAX_WORD];

// Return a pointer to the new node
return new;
Comment Agreement

Comments should be accurate and agree with the code they describe. If you modify some existing code, you should ensure that the surrounding comments are still accurate.

Trailing Comments

Do not use trailing comments (i.e., comments which appear at the end of a line containing other code) unless the comment is very short and describes only the code that it appears next to.

Libraries

Libraries

Unless a task specifies otherwise, you are only permitted to call your own functions and functions from the standard C library.

The standard C library includes functions declared in assert.h, ctype.h, math.h, stdbool.h, stdio.h, stdlib.h and string.h.

You are not permitted to use functions such as read, write and open defined in unistd.h. They are not part of the standard C library.

Note: the math.h library is part of the standard C library and are you are allowed to use it. gcc and clang require you to compile with the -lm option to use the math library.

External Programs

Unless a task specifies otherwise, you are not permitted to run other programs from your C program, using the function system or otherwise.

Warnings, Errors & Undefined Behaviour

Compile-time Warnings

Your code should not generate any warnings or errors when compiled with gcc -Wall -Werror.

You should assume warnings from gcc are errors that need to be fixed.

Invalid C, Run-time errors & Undefined Behaviour

Your code should not perform any operation which results in undefined behaviour or a runtime error (when given any input that is valid for the exercise).

For example, you must ensure your code does not access uninitialized variables, overflow numeric variables, access non-existent illegal array elements or dereference NULL pointers.

All of these will heavily penalized in marking.

Invalid C, Run-time errors & Undefined Behaviour

Your code should not perform any operation which results in undefined behaviour or a runtime error (when given any input that is valid for the exercise).

For example, you must ensure your code does not access uninitialized variables, overflow numeric variables, access non-existent illegal array elements or dereference NULL pointers.

All of these will heavily penalized in marking.

Novice programmers need to be very careful that their code is not invalid.

For example, they need to carefully consider whether an expression used to index an array will always result in a valid array index.

Invalid C is a frequent source of serious security exploits - for example attackers often exploit invalid array indices and integer overflows.

Functions

Repeated Code

Do not cut-and-paste code. The same code should not appear twice in your program. Avoid repeating code by defining a function and calling it twice (or more).

Repeated Code

Do not cut-and-paste code. The same code should not appear twice in your program. Avoid repeating code by defining a function and calling it twice (or more).

Repeated code makes your program hard to read and hard to modify.

Function Purpose

Functions should have one clearly defined purpose and should be short.

"Functions should do one thing, and one thing only."

Function Purpose

Functions should have one clearly defined purpose and should be short.

"Functions should do one thing, and one thing only."

If you have a function that performs two smaller tasks in sequence, make a function for each of those smaller tasks and call them from the more abstract function. Aim for your functions to be 30 lines or less (not including comments). If they are long, think of how you can break them up into smaller functions.

Function Prototypes / Function Signatures

Function prototypes must be used for all functions (except main) and must come before all functions, appearing in the same order as the function implementations.

Function signatures must have the return type, function name, and state the argument list.

Function Prototypes / Function Signatures

Function prototypes must be used for all functions (except main) and must come before all functions, appearing in the same order as the function implementations.

Function signatures must have the return type, function name, and state the argument list.

Function prototypes describe functions by telling the compiler (and anyone reading the program):

  • The type of value the function returns
  • The name of the function
  • The type and name of any arguments to the function
Function Comments

Every function must have a comment placed before the function implementation describing the purpose of the function and any side-effects the function has.

Function Comments

Every function must have a comment placed before the function implementation describing the purpose of the function and any side-effects the function has.

// return pointer to last node in list
// NULL is returned if list is empty

struct node *last(struct node *head) {
Function Names

Function names should be descriptive, typically containing multiple words. Function names containing multiple words should be in either snake_case or camelCase. Do not use both approaches in the one program.

When writing functions for ADTs, you should keep the naming consistent by including the name of the ADT at the beginning. Names of ADT interface functions should be in PascalCase (this rule applies from 21T3 onwards).

Function Arguments

Argument names in a function prototype should match the names used in the function implementation and should always be used in the prototype.

Long function argument lists may be broken over multiple lines.

Functions that do not take any arguments should use (void) instead of an empty argument list.

Function Arguments

Argument names in a function prototype should match the names used in the function implementation and should always be used in the prototype.

Long function argument lists may be broken over multiple lines.

Functions that do not take any arguments should use (void) instead of an empty argument list.

A space may optionally be used between the function name and argument list.

The return type, function name, and start of the argument list should all be on the same line.

If the argument list is too long to fit on a single line with the function name and return type, it may be split across multiple lines, with the additional lines being indented by 2 stops (8 spaces) or aligned with the first argument in the first line.

Function and argument naming is discussed in the functions section of this document.

Function Arguments

Argument names in a function prototype should match the names used in the function implementation and should always be used in the prototype.

Long function argument lists may be broken over multiple lines.

Functions that do not take any arguments should use (void) instead of an empty argument list.

// A short function
int aFunctionWithNoArguments(void);

// A short function
int aSimpleFunction(int argumentA, char *argumentB);

// A function with a lot of arguments
int aLongFunction(int argumentA, int argumentB, int argumentC, int argumentD,
                  int argumentE, int argumentF);

// Another function with a lot of arguments
int anotherLongFunction(int argumentA, int argumentB, int argumentC,
                        int argumentD, int argumentE, int argumentF);

// Another function with a lot of arguments.
// Note that the closing ');' is on a line by itself to make it clearer where
// the prototype ends.
int anotherLongFunction(
    int argumentA, int argumentB, int argumentC,
    int argumentD, int argumentE, int argumentF
);
Return

Do not overuse return statements.

Have multiple multiple returns only if this makes your function less complex or more readable.

For example, it is acceptable to have an early return statement(s) to handle error conditions or special cases, as well as another return statement for the computation of the function's main result.

Be Careful With These C Features

break

Use break statements sensibly.

break

Use break statements sensibly.

Novice programmers often over-use break statements, producing confusing and difficult to debug code.

Use break only when you are sure it makes your code simpler and more readable.

continue

Use continue statements sensibly.

continue

Use continue statements sensibly.

Novice programmers often misuse and over-use continue statements, producing confusing and difficult to debug code.

It is best not to use continue are until you are a more experienced programmer - and then use it carefully ensuring it makes your code more readable.

switch

Use switch statements sensibly.

switch

Use switch statements sensibly.

The C switch statement can have subtle and confusing behaviour.

When using the switch statement, use break statements to avoid fall-through behaviour, unless fall-through is intended.

Ternary if

Use the ternary if construct ((condition) ? true_expression : false_expression) sensibly.

Ternary if

Use the ternary if construct ((condition) ? true_expression : false_expression) sensibly.

Unlike some other languages, C is a statement-oriented language; this means, for example, an if statement doesn't have an inherent value and can't be used inside an expression.

C does have an operator similar to the if statement, the ternary if, which can be used in expressions.

Novice programmers attempting to use the ternary if typically construct unreadable code and confuse themselves.

Most experienced programmers either avoid the ternary if entirely, or use it in very limited circumstances.

do ... while

Use the do { ... } while (condition); construct sensibly.

do ... while

Use the do { ... } while (condition); construct sensibly.

The do { ... } while (condition); construct is similar to a normal while loop, except the body will always run once before the condition is checked.

Novice programmers often confuse themselves and construct hard-to-debug code using do...while loops.

Many experienced programmers avoid do...while entirely, but they are commonly used in some contexts such as kernel programming by experienced programmers.

Avoid These C Features

Macros

Avoid #define statements with parameters (usually called macros) unless you are experienced with them.

Define a function instead.

Macros

Avoid #define statements with parameters (usually called macros) unless you are experienced with them.

Define a function instead.

Macros introduce a nasty class of bugs due to their implementation as textual pre-processing.

Macros were primarily used for efficiency in the past. This is no longer needed.

Modern C compilers will produce the same code for functions as for macros, without the risk of nasty bugs.

Pointer Arithmetic

Avoid pointer arithmetic. Use array indices instead.

Pointer Arithmetic

Avoid pointer arithmetic. Use array indices instead.

Experienced programmers use pointer arithmetic to produce succinct idiomatic code.

Novice programmers often confuse themselves by trying to use pointer arithmetic. Any code using pointer arithmetic can also be written using array indices. Use array indices until you are an experienced programmer.

Type Cast

Avoid type casts, except when converting between numeric types such as from int to double.

Type Cast

Avoid type casts, except when converting between numeric types such as from int to double.

Safe use of type casts, other than between numeric types, involves a deep knowledge of C semantics.

A need for other uses for type casts should not arise in this course, and other uses for type casts should be avoided.

Type Cast

Avoid type casts, except when converting between numeric types such as from int to double.

int x = 42;
int y = 11;
double f;
f = x / ((double) y); // convert y to a double
Goto
Do not use goto.
Goto
Do not use goto.

Most programmers completely avoid use of the goto statement because it can produce incomprehensible code.

Introductory programming courses always ban use of goto to help beginner programmers learn better coding techniques.

There are very limited circumstances, e.g., writing device drivers or emulators, where goto's use is appropriate, but such circumstances will not arise in this course.

There will be a major style deduction if you use goto.

Comma Operator

Avoid using the comma operator to combine multiple expressions.

Comma Operator

Avoid using the comma operator to combine multiple expressions.

Novice programmers attempting to use the comma operator typically construct unreadable code and confuse themselves.

union

Do not use unions; use structs instead.

union

Do not use unions; use structs instead.

Unions allow a polymorphic datatype to occupy less space than if a struct was used.

No need for this should arise in this course.

Unions introduce a nasty class of bugs which are hard to debug and are best avoided by novice programmers.

Unions are also sometimes useful for embedded programming and for building language implementations.