ndn-cxx Code Style and Coding Guidelines

Based on

1. Code layout

1.1. The code layout should generally follow the GNU coding standard layout for C, extended it to C++.

  • Do not use tabs for indentation.

  • Indentation spacing is 2 spaces.

  • Lines should be within a reasonable range. Lines longer than 100 columns should generally be avoided.

1.2. Whitespace

  • Conventional operators (if, for, while, and others) should be surrounded by a space character.

  • Commas should be followed by a white space.

  • Semicolons in for statments should be followed by a space character.

Examples:

a = (b + c) * d;  // NOT: a=(b+c)*d

while (true) {    // NOT: while(true)
  ...

doSomething(a, b, c, d);    // NOT: doSomething(a,b,c,d);

for (i = 0; i < 10; i++) {  // NOT: for(i=0;i<10;i++){
  ...

1.3. Namespaces should have the following form:

namespace example {

code;
moreCode;

} // namespace example

Note that code inside namespace is not indented. Avoid the following:

// NOT
//
// namespace example {
//
//   code;
//   moreCode;
//
// } // namespace example

1.4. Class declarations should have the following form:

class SomeClass : public BaseClass
{
public:
  ... <public methods> ...

protected:
  ... <protected methods> ...

private:
  ... <private methods> ...

public:
  ... <public data> ...

protected:
  ... <protected data> ...

private:
  ... <private data> ...
};

public, protected, private may be repeated several times without interleaving (e.g., public, public, public, private, private) if this improves readability of the code.

Nested classes can be defined in appropriate visibility section, either in methods block, data block, or in a separate section (depending which one provides better code readability).

1.5. Method and function definitions should have the following form:

void
someMethod()
{
  ...
}

void
SomeClass::someMethod()
{
  ...
}

1.6. The if-else class of statements should have the following form:

if (condition) {
  statements;
}

if (condition) {
  statements;
}
else {
  statements;
}

if (condition) {
  statements;
}
else if (condition) {
  statements;
}
else {
  statements;
}

1.7. A for statement should have the following form:

for (initialization; condition; update) {
  statements;
}

An empty for statement should have the following form:

for (initialization; condition; update)
  ;

This emphasizes the fact that the for statement is empty and makes it obvious for the reader that this is intentional. Empty loops should be avoided however.

1.8. A while statement should have the following form:

while (condition) {
  statements;
}

1.9. A do-while statement should have the following form:

do {
  statements;
} while (condition);

1.10. A switch statement should have the following form:

switch (condition) {
  case ABC:        // 2 space indent
    statements;    // 4 space indent
    NDN_CXX_FALLTHROUGH;

  case DEF:
    statements;
    break;

  case XYZ: {
    statements;
    break;
  }

  default:
    statements;
    break;
}

When curly braces are used inside a case block, the braces must cover the entire case block.

switch (condition) {
  // Correct style
  case A0: {
    statements;
    break;
  }

  // Correct style
  case A1: {
    statements;
    NDN_CXX_FALLTHROUGH;
  }

  // Incorrect style: braces should cover the entire case block
  case B: {
    statements;
  }
  statements;
  break;

  default:
    break;
}

The following style is still allowed when none of the case blocks has curly braces.

switch (condition) {
case ABC:        // no indent
  statements;    // 2 space indent
  NDN_CXX_FALLTHROUGH;

case DEF:
  statements;
  break;

default:
  statements;
  break;
}

The NDN_CXX_FALLTHROUGH annotation must be included whenever there is a case without a break statement. Leaving the break out is a common error, and it must be made clear that it is intentional when it is not there. Moreover, modern compilers will warn when a case that falls through is not explicitly annotated.

1.11. A try-catch statement should have the following form:

try {
  statements;
}
catch (const Exception& exception) {
  statements;
}

1.12. The incompleteness of split lines must be made obvious.

totalSum = a + b + c +
           d + e;
function(param1, param2,
         param3);
for (int tableNo = 0; tableNo < nTables;
     tableNo += tableStep) {
  ...
}

Split lines occur when a statement exceeds the column limit given in rule 1.1. It is difficult to give rigid rules for how lines should be split, but the examples above should give a general hint. In general:

  • Break after a comma.

  • Break after an operator.

  • Align the new line with the beginning of the expression on the previous line.

Exceptions:

  • The following is standard practice with operator<<:

    std::cout << "Something here "
              << "Something there" << std::endl;
    

1.13. When class variables need to be initialized in the constructor, the initialization should take the following form:

SomeClass::SomeClass(int value, const std::string& string)
  : m_value(value)
  , m_string(string)
  ...
{
}

Each initialization should be put on a separate line, starting either with the colon for the first initialization or with comma for all subsequent initializations.

1.14. A range-based for statement should have the following form:

for (T i : range) {
  statements;
}

1.15. A lambda expression should have the following form:

[&capture1, capture2] (T1 arg1, T2 arg2) {
  statements;
}

[&capture1, capture2] (T1 arg1, T2 arg2) mutable {
  statements;
}

[this] (T arg) {
  statements;
}

[&] (T arg) {
  statements;
}

[=] (T arg) {
  statements;
}

If the lambda has no parameters, () should be omitted.

[&capture1, capture2] {
  statements;
}

Trailing return types should be omitted whenever possible. Add it only when the compiler cannot deduce the return type automatically, or when it improves readability. Note that () is required by the C++ standard when mutable or a trailing return type is used.

[] (T arg) -> int {
  statements;
}

[] () -> int {
  statements;
}

If the function body has only one line, and the whole lambda expression can fit in one line, the following form is also acceptable:

[&capture1, capture2] (T1 arg1, T2 arg2) { statement; }

A no-op lambda can be written in a more compact form:

[]{}

1.16. List initialization should have the following form:

T object{arg1, arg2};

T{arg1, arg2};

new T{arg1, arg2};

return {arg1, arg2};

function({arg1, arg2}, otherArgument);

object[{arg1, arg2}];

T({arg1, arg2})

T object = {arg1, arg2};

class Class
{
private:
  T m_member{arg1, arg2};
  static T s_member = {arg1, arg2};
};

An empty braced-init-list is written as {}. For example:

T object{};
T object = {};

2. Naming Conventions

2.1. C++ header files should have the extension .hpp. Source files should have the extension .cpp

File names should be all lower case. If the class name is a composite of several words, each word in a file name should be separated with a dash (-). A class should be declared in a header file and defined in a source file where the name of the files match the name of the class.

my-class.hpp, my-class.cpp

2.2. Names representing types must be written in English in mixed case starting with upper case.

class MyClass;
class Line;
class SavingsAccount;

2.3. Variable names must be written in English in mixed case starting with lower case.

MyClass myClass;
Line line;
SavingsAccount savingsAccount;
int theAnswerToLifeTheUniverseAndEverything;

2.4. Named constants (including enumeration values) must be all uppercase using underscore to separate words.

const int MAX_ITERATIONS = 25;
const std::string COLOR_RED = "red";
static const double PI = 3.14;

In some cases, it is a better (or is the only way for complex constants in header-only classes) to implement the value as a method.

static int          // declare constexpr if possible
getMaxIterations()
{
  return 25;
}

2.5. Names representing methods or functions must be commands starting with a verb and written in mixed case starting with lower case.

std::string
getName()
{
  ...
}

double
computeTotalWidth()
{
  ...
}

2.6. Names representing namespaces should be all lowercase.

namespace model {
namespace analyzer {

...

} // namespace analyzer
} // namespace model

2.7. Names representing generic template types should be a single uppercase letter.

template<class T> ...
template<class C, class D> ...

However, when a template parameter represents a certain concept and is expected to have a certain interface, the name should be explicitly spelled out.

template<class InputIterator> ...
template<class Packet> ...

2.8. Abbreviations and acronyms must not be uppercase when used as name.

exportHtmlSource(); // NOT: exportHTMLSource();
openDvdPlayer();    // NOT: openDVDPlayer();

2.9. Global variables should have g_ prefix

g_mainWindow.open();
g_applicationContext.getName();

In general, the use of global variables should be avoided. Consider using singleton objects instead.

2.10. All non-static data members of a class should be prefixed with m_ unless they are public. Similarly, non-public static data members should be prefixed with s_.

class SomeClass
{
private:
  int m_length;

  static std::string s_name;
};

2.11. Variables with a large scope should have long (explicit) names, variables with a small scope can have short names.

Scratch variables used for temporary storage or indices are best kept short. A programmer reading such variables should be able to assume that its value is not used outside of a few lines of code. Common scratch variables for integers are i, j, k, m, n and for characters c and d.

2.12. The name of the object is implicit, and should be avoided in a method name.

line.getLength(); // NOT: line.getLineLength();

The latter seems natural in the class declaration, but proves superfluous in use, as shown in the example.

2.13. The terms get/set must be used where an attribute is accessed directly.

employee.getName();
employee.setName(name);

matrix.getElement(2, 4);
matrix.setElement(2, 4, value);

2.14. The term compute can be used in methods where something is computed.

valueSet.computeAverage();
matrix.computeInverse()

Give the reader the immediate clue that this is a potentially time-consuming operation, and if used repeatedly, he might consider caching the result. Consistent use of the term enhances readability.

2.15. The term find can be used in methods where something is looked up.

vertex.findNearestVertex();
matrix.findMinElement();

Give the reader the immediate clue that this is a simple look up method with a minimum of computations involved. Consistent use of the term enhances readability.

2.16. Plural form should be used on names representing a collection of objects.

vector<Point> points;
int values[];

Enhances readability since the name gives the user an immediate clue of the type of the variable and the operations that can be performed on its elements.

2.17. The prefix n should be used for variables representing a number of objects.

nPoints, nLines

The notation is taken from mathematics where it is an established convention for indicating a number of objects.

2.18. The suffix Num or No should be used for variables representing an entity number.

tableNum, tableNo, employeeNum, employeeNo

2.19. The prefix is, has, need, or similar should be used for boolean variables and methods.

isSet, isVisible, isFinished, isFound, isOpen
needToConvert, needToFinish

2.20. Complement names must be used for complement operations, reducing complexity by symmetry.

get/set, add/remove, create/destroy, start/stop, insert/erase,
increment/decrement, old/new, begin/end, first/last, up/down, min/max,
next/previous (and commonly used next/prev), open/close, show/hide,
suspend/resume, etc.

The pair insert/erase is preferred. insert/delete can also be used if it does not conflict with the delete keyword of C++.

2.21. Variable names should not include reference to variable type (do not use Hungarian notation).

Line* line; // NOT: Line* pLine;
            // NOT: Line* linePtr;

size_t nPoints; // NOT lnPoints

char* name; // NOT szName

2.22. Negated boolean variable names should be avoided.

bool isError; // NOT: isNoError
bool isFound; // NOT: isNotFound

2.23. Enumeration constants recommended to prefix with a common type name.

enum Color {
  COLOR_RED,
  COLOR_GREEN,
  COLOR_BLUE
};

2.24. Exceptions can be suffixed with either Exception (e.g., SecurityException) or Error (e.g., SecurityError).

The recommended method is to declare an exception class Exception or Error as a nested type inside the class from which the exception is thrown. For example, when defining a class Foo that can throw errors, one can write the following:

#include <stdexcept>

class Foo
{
public:
  class Error : public std::runtime_error
  {
  public:
    // You can inherit constructors from std::runtime_error like this:
    using std::runtime_error::runtime_error;

    // Additional constructors, if desired, can be declared as usual:
    Error(const std::string& what, const std::exception& inner)
      : std::runtime_error(what + ": " + inner.what())
    {
    }
  };
};

In addition to that, if class Foo is a base class or interface for some class hierarchy, then child classes should should define their own Error or Exception classes that are inherited from the parent’s Error class.

2.25. Functions (methods returning something) should be named after what they return and procedures (void methods) after what they do.

Increase readability. Makes it clear what the unit should do and especially all the things it is not supposed to do. This again makes it easier to keep the code clean of side effects.

3. Miscellaneous

3.1. Exceptions can be used in the code, but should be used only in exceptional cases and not in the primary processing path.

3.2. Header files must contain an include guard.

For example, a header file named module/class-name.hpp or src/module/class-name.hpp should have a header guard in the following form:

#ifndef APP_MODULE_CLASS_NAME_HPP
#define APP_MODULE_CLASS_NAME_HPP
...
#endif // APP_MODULE_CLASS_NAME_HPP

The macro name should reflect the path of the header file relative to the root of the source tree, in order to prevent naming conflicts. The header guard should be prefixed with the application/library name to avoid conflicts with other packages and libraries.

3.3. Include directives for system headers and other external libraries should use <angle brackets>. Header files in the same source code repository should be included using "quotes".

#include "ndn-cxx/util/random.hpp"

#include <string>
#include <boost/lexical_cast.hpp>

All of a project’s header files should be included with their path relative to the project’s source directory. The use of UNIX directory shortcuts . (the current directory) and .. (the parent directory) is discouraged.

3.4. Include statements should be grouped. Same-project headers should be included first. Leave an empty line between groups of include statements. Sort alphabetically within a group. For example, the include section of ndn-cxx/foo/bar.cpp may look like this:

#include "ndn-cxx/impl/pending-interest.hpp"
#include "ndn-cxx/util/random.hpp"

#include <cstdlib>
#include <fstream>
#include <iomanip>

#include <boost/lexical_cast.hpp>
#include <boost/regex.hpp>

3.5. Definitions that are local to only one .cpp file should be declared inside that file and be placed in an unnamed namespace or declared static.

3.6. Implicit conversion is generally allowed.

Implicit conversion between integer and floating point numbers can cause problems and should be avoided.

Implicit conversion in constructors that can be called with a single argument is usually undesirable. Therefore, all single-argument constructors should be marked explicit, unless implicit conversion is desirable. In that case, a comment should document the reason for this. As an exception, copy and move constructors should not be explicit, since they do not perform type conversion. Constructors that cannot be called with a single argument may omit explicit. Constructors that take a single std::initializer_list parameter should also omit explicit, in order to support copy-initialization.

Avoid C-style casts. Use static_cast, dynamic_cast, const_cast, reinterpret_cast, or bit_cast instead where appropriate. Use static_pointer_cast, dynamic_pointer_cast, or const_pointer_cast when dealing with shared_ptr.

3.7. Variables should be initialized where they are declared.

This ensures that variables are valid at any time. Sometimes it is impossible to initialize a variable to a valid value where it is declared.

int x, y, z;
getCenter(&x, &y, &z);

In these cases it should be left uninitialized rather than initialized to some phony value.

3.8. In most cases, class data members should not be declared public.

Public data members violate the concepts of information hiding and encapsulation. Use private variables and public accessor methods instead.

Exceptions to this rule:

  • When the class is essentially a passive data structure with no or minimal behavior (equivalent to a C struct, also known as POD type). In this case, all fields should be public and the keyword struct should be used instead of class.

  • When the class is used only inside the compilation unit, e.g., when implementing pImpl idiom (aka Bridge pattern) or similar cases.

3.9. C++ pointers and references should have their reference symbol next to the type rather than to the name.

float* x; // NOT: float *x;
int& y;   // NOT: int &y;

3.10. Implicit test for 0 should not be used other than for boolean variables and pointers.

if (nLines != 0)    // NOT: if (nLines)

int* ptr = ...
if (ptr)            // OK
if (ptr != nullptr) // also OK

3.11. (removed)

3.12. Loop variables should be initialized immediately before the loop.

bool isDone = false;   // NOT: bool isDone = false;
while (!isDone) {      //      // other stuff
  ...                  //      while (!isDone) {
}                      //        ...
                       //      }

3.13. The form while (true) should be used for infinite loops.

while (true) {
  ...
}

// NOT:
for (;;) { // NO!
  ...
}
while (1) { // NO!
  ...
}

3.14. Complex conditional expressions must be avoided. Introduce temporary boolean variables instead.

bool isFinished = (elementNo < 0) || (elementNo > maxElement);
bool isRepeatedEntry = elementNo == lastElement;
if (isFinished || isRepeatedEntry) {
  ...
}

// NOT:
// if ((elementNo < 0) || (elementNo > maxElement) || elementNo == lastElement) {
//  ...
// }

By assigning boolean variables to expressions, the program gets automatic documentation. The construction will be easier to read, debug, and maintain.

3.15. The conditional should be put on a separate line.

if (isDone)         // NOT: if (isDone) doCleanup();
  doCleanup();

This is for debugging purposes. When writing on a single line, it is not apparent whether the test is really true or not.

3.16. Assignment statements in conditionals must be avoided.

File* fileHandle = open(fileName, "w");
if (!fileHandle) {
  ...
}

// NOT
// if (!(fileHandle = open(fileName, "w"))) {
//  ..
// }

3.17. The use of magic numbers in the code should be avoided. Numbers other than 0 and 1 should be considered declared as named constants instead.

If the number does not have an obvious meaning by itself, the readability is enhanced by introducing a named constant instead. A different approach is to introduce a method from which the constant can be accessed.

3.18. Floating point literals should always be written with a decimal point, at least one decimal, and without omitting 0 before the decimal point.

double total = 0.0;     // NOT: double total = 0;
double someValue = 0.1; // NOT double someValue = .1;
double speed = 3.0e8;   // NOT: double speed = 3e8;
double sum;
...
sum = (a + b) * 10.0;

3.19. goto should not be used.

goto statements violate the idea of structured code. Only in very few cases (for instance, breaking out of deeply nested structures) should goto be considered, and only if the alternative structured counterpart is proven to be less readable.

3.20. nullptr should be used to represent a null pointer, instead of “0” or “NULL”.

3.21. Logical units within a block should be separated by one blank line.

Matrix4x4 matrix = new Matrix4x4();

double cosAngle = Math.cos(angle);
double sinAngle = Math.sin(angle);

matrix.setElement(1, 1, cosAngle);
matrix.setElement(1, 2, sinAngle);
matrix.setElement(2, 1, -sinAngle);
matrix.setElement(2, 2, cosAngle);

multiply(matrix);

Enhance readability by introducing white space between logical units of a block.

3.22. Variables in declarations can be left aligned.

AsciiFile* file;
int        nPoints;
float      x, y;

Enhance readability. The variables are easier to spot from the types by alignment.

3.23. Use alignment wherever it enhances readability.

value = (potential        * oilDensity)   / constant1 +
        (depth            * waterDensity) / constant2 +
        (zCoordinateValue * gasDensity)   / constant3;

minPosition =     computeDistance(min,     x, y, z);
averagePosition = computeDistance(average, x, y, z);

There are a number of places in the code where white space can be included to enhance readability even if this violates common guidelines. Many of these cases have to do with code alignment. General guidelines on code alignment are difficult to give, but the examples above should give a general clue.

3.24. All comments should be written in English.

In an international environment, English is the preferred language.

3.25. Use // for all comments, including multi-line comments.

// Comment spanning
// more than one line.

Since multilevel C-commenting is not supported, using // comments ensure that it is always possible to comment out entire sections of a file using /* */ for debugging purposes etc.

There should be a space between the // and the actual comment, and comments should always start with an upper case letter and end with a period.

However, method and class documentation comments should use /** */ style for Doxygen, JavaDoc and JSDoc. License boilerplate should use /* */ style.

3.26. Comments should be included relative to their position in the code.

while (true) {
  // Do something
  something();
}

// NOT:
while (true) {
// Do something
  something();
}

This is to avoid that the comments break the logical structure of the program.

3.27. Use BOOST_ASSERT and BOOST_ASSERT_MSG for runtime assertions.

int x = 1;
int y = 2;
int z = x + y;
BOOST_ASSERT(z - y == x);

The expression passed to BOOST_ASSERT must not have side effects, because it may not be evaluated in release builds.

3.28. Use static_assert for compile-time assertions.

class BaseClass
{
};

class DerivedClass : public BaseClass
{
};

static_assert(std::is_base_of<BaseClass, DerivedClass>::value,
              "DerivedClass must inherit from BaseClass");

3.29. The auto type specifier may be used for local variables if a human reader can easily deduce the actual type, or if it makes the code safer.

std::vector<int> intVector;
auto i = intVector.find(4); // OK

auto stringSet = std::make_shared<std::set<std::string>>(); // OK

std::vector<std::string> strings;
for (const auto& str : strings) { // OK, iterating over the elements of a container
  std::cout << str;
}

obj.onEvent([] (auto&&...) { std::cout << "hi!"; }); // OK, unused lambda parameters

auto x = foo(); // BAD unless foo() is declared nearby or has a well-known prototype

3.30. Use the override or final specifier when overriding a virtual member function or a virtual destructor.

virtual must not be used along with final so that the compiler can generate an error when a final function does not override.

virtual should not be used along with override for consistency with final.

class Stream
{
public:
  virtual void
  open();
};

class InputStream : public Stream
{
public:
  void
  open() override;
};

class Console : public InputStream
{
public:
  void
  open() final;
};

3.31. The recommended way to throw an exception derived from std::exception is to use NDN_THROW or one of the other NDN_THROW_* macros.

Exceptions thrown using these macros will be augmented with additional diagnostic information, including the file name, line number, and function name from which the exception was thrown.

The extended diagnostic information contained in the exception can be printed with boost::diagnostic_information().

#include <boost/exception/diagnostic_information.hpp>
#include <iostream>

try {
  operationThatMayThrow();
}
catch (const std::exception& e) {
  std::cerr << boost::diagnostic_information(e);
}