Google C++ Style Guide

原文地址：http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml

Google C++ Style Guide

Revision 3.260

Benjy Weinberger
Craig Silverstein
Gregory Eitzmann
Mark Mentovai
Tashana Landray

Each style point has a summary for whichadditional information is available by toggling the accompanying arrow buttonthat looks this way: ▽. You may toggle all summaries with thebig arrow button:

▽ Toggle all summaries

Table of Contents

Header Files

The #define Guard Forward Declarations Inline Functions The -inl.h Files Function Parameter Ordering Names and Order of Includes

Scoping

Namespaces Nested Classes Nonmember, Static Member, and Global Functions Local Variables Static and Global Variables

Classes

Doing Work in Constructors Default Constructors Explicit Constructors Copy Constructors Structs vs. Classes Inheritance Multiple InheritanceInterfaces Operator Overloading Access Control Declaration Order Write Short Functions

Google-Specific Magic

Smart Pointers cpplint

Other C++ Features

Reference Arguments Function Overloading Default Arguments Variable-Length Arrays and alloca() Friends ExceptionsRun-Time Type Information (RTTI) Casting Streams Preincrement and Predecrement Use of const Integer Types 64-bit PortabilityPreprocessor Macros 0 and nullptr/NULL sizeof auto Brace Initialization Boost C++11 unique_ptr

Naming

General Naming Rules File Names Type Names Variable Names Constant Names Function Names Namespace Names Enumerator Names Macro NamesExceptions to Naming Rules

Comments

Comment Style File Comments Class Comments Function Comments Variable Comments Implementation Comments Punctuation, Spelling and GrammarTODO Comments Deprecation Comments

Formatting

Line Length Non-ASCII Characters Spaces vs. Tabs Function Declarations and Definitions Function Calls Braced Initializer Lists ConditionalsLoops and Switch Statements Pointer and Reference Expressions Boolean Expressions Return Values Variable and Array InitializationPreprocessor Directives Class Format Constructor Initializer Lists Namespace Formatting Horizontal Whitespace Vertical Whitespace

Exceptions to the Rules

Existing Non-conformant Code Windows Code

Important Note

Displaying Hidden Details in this Guide

link▽

This style guide contains many detailsthat are initially hidden from view. They are marked by the triangle icon,which you see here on your left. Click it now. You should see"Hooray" appear below.

Hooray! Now you know you can expandpoints to get more details. Alternatively, there's an "expand all" atthe top of this document.

Background

C++ is the main development languageused by many of Google's open-source projects. As every C++ programmer knows,the language has many powerful features, but this power brings with itcomplexity, which in turn can make code more bug-prone and harder to read andmaintain.

The goal of this guide is to manage thiscomplexity by describing in detail the dos and don'ts of writing C++ code.These rules exist to keep the code base manageable while still allowing codersto use C++ language features productively.

Style, also known as readability, is what wecall the conventions that govern our C++ code. The term Style is a bit of amisnomer, since these conventions cover far more than just source fileformatting.

One way in which we keep the code basemanageable is by enforcing consistency. It is very important thatany programmer be able to look at another's code and quickly understand it.Maintaining a uniform style and following conventions means that we can moreeasily use "pattern-matching" to infer what various symbols are andwhat invariants are true about them. Creating common, required idioms andpatterns makes code much easier to understand. In some cases there might begood arguments for changing certain style rules, but we nonetheless keep thingsas they are in order to preserve consistency.

Another issue this guide addresses isthat of C++ feature bloat. C++ is a huge language with many advanced features.In some cases we constrain, or even ban, use of certain features. We do this tokeep code simple and to avoid the various common errors and problems that thesefeatures can cause. This guide lists these features and explains why their useis restricted.

Open-source projects developed by Googleconform to the requirements in this guide.

Note that this guide is not a C++tutorial: we assume that the reader is familiar with the language.

Header Files

In general, every .cc file should have an associated .h file. There are some common exceptions, such asunittests and small .cc files containing just a main() function.

Correct use of header files can make ahuge difference to the readability, size and performance of your code.

The following rules will guide youthrough the various pitfalls of using header files.

The #define Guard

link▽

All header files should have #define guards to prevent multipleinclusion. The format of the symbol name should be <PROJECT>_<PATH>_<FILE>_H_.

To guarantee uniqueness, they should bebased on the full path in a project's source tree. For example, the file foo/src/bar/baz.h in project foo should have the following guard:

#ifndef FOO_BAR_BAZ_H_

#define FOO_BAR_BAZ_H_

 

...

 

#endif // FOO_BAR_BAZ_H_

Forward Declarations

link▽

You may forward declare ordinary classesin order to avoid unnecessary #includes.

Definition:A "forward declaration" is adeclaration of a class, function, or template without an associated definition. #include lines can often be replaced withforward declarations of whatever symbols are actually used by the client code.

Pros:

• Unnecessary #includes force the compiler to open more files and process more input.
• They can also force your code to be recompiled more often, due to changes in the header.

Cons:

• It can be difficult to determine the correct form of a forward declaration in the presence of features like templates, typedefs, default parameters, and using declarations.
• It can be difficult to determine whether a forward declaration or a full #include is needed for a given piece of code, particularly when implicit conversion operations are involved. In extreme cases, replacing an #include with a forward declaration can silently change the meaning of code.
• Forward declaring multiple symbols from a header can be more verbose than simply #includeing the header.
• Forward declarations of functions and templates can prevent the header owners from making otherwise-compatible changes to their APIs; for example, widening a parameter type, or adding a template parameter with a default value.
• Forward declaring symbols from namespace std:: usually yields undefined behavior.
• Structuring code to enable forward declarations (e.g. using pointer members instead of object members) can make the code slower and more complex.
• The practical efficiency benefits of forward declarations are unproven.

Decision:

• When using a function declared in a header file, always #include that header.
• When using a class template, prefer to #include its header file.
• When using an ordinary class, relying on a forward declaration is OK, but be wary of situations where a forward declaration may be insufficient or incorrect; when in doubt, just #include the appropriate header.
• Do not replace data members with pointers just to avoid an #include.

Always #include the file that actually provides thedeclarations/definitions you need; do not rely on the symbol being brought intransitively via headers not directly included. One exception is that myfile.cc may rely on #includes and forward declarations from itscorresponding header file myfile.h.

Inline Functions

link▽

Define functions inline only when theyare small, say, 10 lines or less.

Definition:You can declare functions in a way thatallows the compiler to expand them inline rather than calling them through theusual function call mechanism.

Pros:Inlining a function can generate moreefficient object code, as long as the inlined function is small. Feel free toinline accessors and mutators, and other short, performance-critical functions.

Cons:Overuse of inlining can actually makeprograms slower. Depending on a function's size, inlining it can cause the codesize to increase or decrease. Inlining a very small accessor function willusually decrease code size while inlining a very large function candramatically increase code size. On modern processors smaller code usually runsfaster due to better use of the instruction cache.

Decision:

A decent rule of thumb is to not inlinea function if it is more than 10 lines long. Beware of destructors, which areoften longer than they appear because of implicit member- and base-destructorcalls!

Another useful rule of thumb: it'stypically not cost effective to inline functions with loops or switchstatements (unless, in the common case, the loop or switch statement is neverexecuted).

It is important to know that functionsare not always inlined even if they are declared as such; for example, virtualand recursive functions are not normally inlined. Usually recursive functionsshould not be inline. The main reason for making a virtual function inline isto place its definition in the class, either for convenience or to document itsbehavior, e.g., for accessors and mutators.

The -inl.h Files

link▽

You may use file names with a -inl.h suffix to define complex inlinefunctions when needed.

The definition of an inline functionneeds to be in a header file, so that the compiler has the definition availablefor inlining at the call sites. However, implementation code properly belongsin .cc files, and we do not like to havemuch actual code in .h files unless there is areadability or performance advantage.

If an inline function definition isshort, with very little, if any, logic in it, you should put the code in your .h file. For example, accessors and mutators shouldcertainly be inside a class definition. More complex inline functions may alsobe put in a .h file for the convenience of theimplementer and callers, though if this makes the .h file too unwieldy you can instead put that code ina separate -inl.h file. This separates theimplementation from the class definition, while still allowing theimplementation to be included where necessary.

Another use of -inl.h files is for definitions offunction templates. This can be used to keep your template definitions easy toread.

Do not forget that a -inl.h file requires a #define guard justlike any other header file.

Function Parameter Ordering

link▽

When defining a function, parameterorder is: inputs, then outputs.

Parameters to C/C++ functions are eitherinput to the function, output from the function, or both. Input parameters areusually values or const references, while output andinput/output parameters will be non-const pointers.When ordering function parameters, put all input-only parameters before anyoutput parameters. In particular, do not add new parameters to the end of thefunction just because they are new; place new input-only parameters before theoutput parameters.

This is not a hard-and-fast rule. Parametersthat are both input and output (often classes/structs) muddy the waters, and,as always, consistency with related functions may require you to bend the rule.

Names and Order of Includes

link▽

Use standard order for readability andto avoid hidden dependencies: C library, C++ library, other libraries' .h, your project's .h.

All of a project's header files shouldbe listed as descendants of the project's source directory without use of UNIXdirectory shortcuts . (the current directory) or .. (the parent directory). For example, google-awesome-project/src/base/logging.h should be included as

#include "base/logging.h"

In dir/foo.cc or dir/foo_test.cc, whose main purpose is to implement or test the stuff in dir2/foo2.h, order your includes as follows:

1. dir2/foo2.h (preferred location — see details below).
2. C system files.
3. C++ system files.
4. Other libraries' .h files.
5. Your project's .h files.

With the preferred ordering, if dir2/foo2.h omits any necessary includes, the build of dir/foo.cc or dir/foo_test.cc will break. Thus, this ruleensures that build breaks show up first for the people working on these files,not for innocent people in other packages.

dir/foo.cc and dir2/foo2.h are often in the same directory (e.g. base/basictypes_test.cc and base/basictypes.h), but can be in different directoriestoo.

Within each section the includes shouldbe ordered alphabetically. Note that older code might not conform to this ruleand should be fixed when convenient.

For example, the includes in google-awesome-project/src/foo/internal/fooserver.cc might look like this:

#include"foo/public/fooserver.h"  //Preferred location.

 

#include <sys/types.h>

#include <unistd.h>

#include <hash_map>

#include <vector>

 

#include "base/basictypes.h"

#include"base/commandlineflags.h"

#include "foo/public/bar.h"

Scoping

Namespaces

link▽

Unnamed namespaces in .cc files are encouraged. With namednamespaces, choose the name based on the project, and possibly its path. Do notuse a using-directive.

Definition:Namespaces subdivide the global scopeinto distinct, named scopes, and so are useful for preventing name collisionsin the global scope.

Pros:

Namespaces provide a (hierarchical) axisof naming, in addition to the (also hierarchical) name axis provided byclasses.

For example, if two different projectshave a class Foo in the global scope, these symbolsmay collide at compile time or at runtime. If each project places their code ina namespace, project1::Foo and project2::Foo are now distinct symbols that do not collide.

Cons:

Namespaces can be confusing, becausethey provide an additional (hierarchical) axis of naming, in addition to the(also hierarchical) name axis provided by classes.

Use of unnamed namespaces in headerfiles can easily cause violations of the C++ One Definition Rule (ODR).

Decision:

Use namespaces according to the policydescribed below. Terminate namespaces with comments as shown in the givenexamples.

Unnamed Namespaces

• Unnamed namespaces are allowed and even encouraged in .cc files, to avoid runtime naming conflicts:

·        namespace{                           // This is ina .cc file.

·         

·        //The content of a namespace is not indented

·        enum{ kUnused, kEOF, kError };       //Commonly used tokens.

·        boolAtEof() { return pos_ == kEOF; }  // Usesour namespace's EOF.

·         

}  //namespace

However, file-scope declarations thatare associated with a particular class may be declared in that class as types,static data members or static member functions rather than as members of anunnamed namespace.

• Do not use unnamed namespaces in .h files.

Named Namespaces

Named namespaces should be used asfollows:

• Namespaces wrap the entire source file after includes, gflags definitions/declarations, and forward declarations of classes from other namespaces:

·        //In the .h file

·        namespacemynamespace {

·         

·        //All declarations are within the namespace scope.

·        //Notice the lack of indentation.

·        classMyClass {

·         public:

·          ...

·          void Foo();

·        };

·         

}  //namespace mynamespace

// In the .cc file

namespace mynamespace {

 

// Definition of functions is within scopeof the namespace.

void MyClass::Foo() {

  ...

}

 

}  //namespace mynamespace

The typical .cc file might have more complexdetail, including the need to reference classes in other namespaces.

#include "a.h"

 

DEFINE_bool(someflag, false, "dummyflag");

 

class C;  // Forward declaration of class C in theglobal namespace.

namespace a { class A; }  // Forward declaration of a::A.

 

namespace b {

 

...code for b...         // Code goes against the left margin.

 

}  //namespace b

• Do not declare anything in namespace std, not even forward declarations of standard library classes. Declaring entities in namespace std is undefined behavior, i.e., not portable. To declare entities from the standard library, include the appropriate header file.
• You may not use a using-directive to make all names from a namespace available.

·        //Forbidden -- This pollutes the namespace.

using namespace foo;

• You may use a using-declaration anywhere in a .cc file, and in functions, methods or classes in .h files.

·        //OK in .cc files.

·        //Must be in a function, method or class in .h files.

using ::foo::bar;

• Namespace aliases are allowed anywhere in a .cc file, anywhere inside the named namespace that wraps an entire .h file, and in functions and methods.

·        //Shorten access to some commonly used names in .cc files.

·        namespacefbz = ::foo::bar::baz;

·         

·        //Shorten access to some commonly used names (in a .h file).

·        namespacelibrarian {

·        //The following alias is available to all files including

·        //this header (in namespace librarian):

·        //alias names should therefore be chosen consistently

·        //within a project.

·        namespacepd_s = ::pipeline_diagnostics::sidetable;

·         

·        inlinevoid my_inline_function() {

·          // namespace alias local to a function (ormethod).

·          namespace fbz = ::foo::bar::baz;

·          ...

·        }

}  //namespace librarian

Note that an alias in a .h file isvisible to everyone #including that file, so public headers (those availableoutside a project) and headers transitively #included by them, should avoiddefining aliases, as part of the general goal of keeping public APIs as smallas possible.

Nested Classes

link▽

Although you may use public nestedclasses when they are part of an interface, consider a namespace tokeep declarations out of the global scope.

Definition:A class can define another class withinit; this is also called a member class.

class Foo {

 

 private:

  //Bar is a member class, nested within Foo.

 class Bar {

   ...

  };

 

};

Pros:This is useful when the nested (ormember) class is only used by the enclosing class; making it a member puts itin the enclosing class scope rather than polluting the outer scope with theclass name. Nested classes can be forward declared within the enclosing classand then defined in the .cc file to avoid including the nestedclass definition in the enclosing class declaration, since the nested classdefinition is usually only relevant to the implementation.

Cons:Nested classes can be forward-declaredonly within the definition of the enclosing class. Thus, any header filemanipulating a Foo::Bar* pointer will have to include thefull class declaration for Foo.

Decision:Do not make nested classes public unlessthey are actually part of the interface, e.g., a class that holds a set ofoptions for some method.

Nonmember, Static Member, and GlobalFunctions

link▽

Prefer nonmember functions within anamespace or static member functions to global functions; use completely globalfunctions rarely.

Pros:Nonmember and static member functionscan be useful in some situations. Putting nonmember functions in a namespace avoidspolluting the global namespace.

Cons:Nonmember and static member functionsmay make more sense as members of a new class, especially if they accessexternal resources or have significant dependencies.

Decision:

Sometimes it is useful, or even necessary,to define a function not bound to a class instance. Such a function can beeither a static member or a nonmember function. Nonmember functions should notdepend on external variables, and should nearly always exist in a namespace.Rather than creating classes only to group static member functions which do notshare static data, use namespaces instead.

Functions defined in the samecompilation unit as production classes may introduce unnecessary coupling andlink-time dependencies when directly called from other compilation units;static member functions are particularly susceptible to this. Considerextracting a new class, or placing the functions in a namespace possibly in aseparate library.

If you must define a nonmember functionand it is only needed in its .cc file, usean unnamed namespace or static linkage (eg static int Foo() {...}) to limit its scope.

Local Variables

link▽

Place a function's variables in thenarrowest scope possible, and initialize variables in the declaration.

C++ allows you to declare variablesanywhere in a function. We encourage you to declare them in as local a scope aspossible, and as close to the first use as possible. This makes it easier forthe reader to find the declaration and see what type the variable is and whatit was initialized to. In particular, initialization should be used instead ofdeclaration and assignment, e.g.

int i;

i = f();      // Bad -- initialization separate fromdeclaration.

int j = g();  // Good -- declaration has initialization.

vector<int> v;

v.push_back(1);  // Prefer initializing using braceinitialization.

v.push_back(2);

vector<int> v = {1, 2};  // Good -- v starts initialized.

Note that gcc implements for (int i = 0; i < 10; ++i) correctly (the scope of i is only the scope of the for loop), so you can then reuse i in another for loop in the same scope. It also correctly scopesdeclarations in if and while statements, e.g.

while (const char* p = strchr(str, '/'))str = p + 1;

There is one caveat: if the variable isan object, its constructor is invoked every time it enters scope and iscreated, and its destructor is invoked every time it goes out of scope.

// Inefficient implementation:

for (int i = 0; i < 1000000; ++i) {

  Foof;  // My ctor and dtor get called1000000 times each.

 f.DoSomething(i);

}

It may be more efficient to declare sucha variable used in a loop outside that loop:

Foo f; // My ctor and dtor get called once each.

for (int i = 0; i < 1000000; ++i) {

 f.DoSomething(i);

}

Static and Global Variables

link▽

Static or global variables of class typeare forbidden: they cause hard-to-find bugs due to indeterminate order ofconstruction and destruction.

Objects with static storage duration,including global variables, static variables, static class member variables,and function static variables, must be Plain Old Data (POD): only ints, chars,floats, or pointers, or arrays/structs of POD.

The order in which class constructorsand initializers for static variables are called is only partially specified inC++ and can even change from build to build, which can cause bugs that aredifficult to find. Therefore in addition to banning globals of class type, we donot allow static POD variables to be initialized with the result of a function,unless that function (such as getenv(), or getpid()) does not itself depend onany other globals.

Likewise, the order in which destructorsare called is defined to be the reverse of the order in which the constructorswere called. Since constructor order is indeterminate, so is destructor order.For example, at program-end time a static variable might have been destroyed,but code still running -- perhaps in another thread -- tries to access it andfails. Or the destructor for a static 'string' variable might be run prior tothe destructor for another variable that contains a reference to that string.

As a result we only allow staticvariables to contain POD data. This rule completely disallows vector (use C arrays instead), or string (use const char []).

If you need a static or global variableof a class type, consider initializing a pointer (which will never be freed),from either your main() function or from pthread_once(). Note that this must bea raw pointer, not a "smart" pointer, since the smart pointer'sdestructor will have the order-of-destructor issue that we are trying to avoid.

Classes

Classes are the fundamental unit of codein C++. Naturally, we use them extensively. This section lists the main dos anddon'ts you should follow when writing a class.

Doing Work in Constructors

link▽

Avoid doing complex initialization inconstructors (in particular, initialization that can fail or that requiresvirtual method calls).

Definition:It is possible to perform initializationin the body of the constructor.

Pros:Convenience in typing. No need to worryabout whether the class has been initialized or not.

Cons:The problems with doing work inconstructors are:

• There is no easy way for constructors to signal errors, short of using exceptions (which are forbidden).
• If the work fails, we now have an object whose initialization code failed, so it may be an indeterminate state.
• If the work calls virtual functions, these calls will not get dispatched to the subclass implementations. Future modification to your class can quietly introduce this problem even if your class is not currently subclassed, causing much confusion.
• If someone creates a global variable of this type (which is against the rules, but still), the constructor code will be called before main(), possibly breaking some implicit assumptions in the constructor code. For instance, gflags will not yet have been initialized.

Decision:Constructors should never call virtualfunctions or attempt to raise non-fatal failures. If your object requiresnon-trivial initialization, consider using a factory function or Init() method.

Default Constructors

link▽

You must define a default constructor ifyour class defines member variables and has no other constructors. Otherwisethe compiler will do it for you, badly.

Definition:The default constructor is called whenwe new a class object with no arguments.It is always called when calling new[] (forarrays).

Pros:Initializing structures by default, tohold "impossible" values, makes debugging much easier.

Cons:Extra work for you, the code writer.

Decision:

If your class defines member variablesand has no other constructors you must define a default constructor (one thattakes no arguments). It should preferably initialize the object in such a waythat its internal state is consistent and valid.

The reason for this is that if you haveno other constructors and do not define a default constructor, the compilerwill generate one for you. This compiler generated constructor may notinitialize your object sensibly.

If your class inherits from an existingclass but you add no new member variables, you are not required to have adefault constructor.

Explicit Constructors

link▽

Use the C++ keyword explicit for constructors with oneargument.

Definition:Normally, if a constructor takes oneargument, it can be used as a conversion. For instance, if you define Foo::Foo(string name) and then pass a string to afunction that expects a Foo, the constructor will be called toconvert the string into a Foo and will pass the Foo to your function for you. This canbe convenient but is also a source of trouble when things get converted and newobjects created without you meaning them to. Declaring a constructor explicit prevents it from being invokedimplicitly as a conversion.

Pros:Avoids undesirable conversions.

Cons:None.

Decision:

We require all single argumentconstructors to be explicit. Always put explicit in front of one-argument constructors in the classdefinition: explicitFoo(string name);

The exception is copy constructors,which, in the rare cases when we allow them, should probably not be explicit. Classes that are intended to betransparent wrappers around other classes are also exceptions. Such exceptions shouldbe clearly marked with comments.

Finally, constructors that take only aninitializer_list may be non-explicit. This is to permit construction of yourtype using the assigment form for brace init lists (i.e.MyType m = {1, 2} ).

Copy Constructors

link▽

Provide a copy constructor andassignment operator only when necessary. Otherwise, disable them with DISALLOW_COPY_AND_ASSIGN.

Definition:The copy constructor and assignmentoperator are used to create copies of objects. The copy constructor isimplicitly invoked by the compiler in some situations, e.g. passing objects byvalue.

Pros:Copy constructors make it easy to copyobjects. STL containers require that all contents be copyable and assignable.Copy constructors can be more efficient than CopyFrom()-style workarounds because they combine construction withcopying, the compiler can elide them in some contexts, and they make it easierto avoid heap allocation.

Cons:Implicit copying of objects in C++ is arich source of bugs and of performance problems. It also reduces readability,as it becomes hard to track which objects are being passed around by value asopposed to by reference, and therefore where changes to an object arereflected.

Decision:

Few classes need to be copyable. Mostshould have neither a copy constructor nor an assignment operator. In manysituations, a pointer or reference will work just as well as a copied value,with better performance. For example, you can pass function parameters byreference or pointer instead of by value, and you can store pointers ratherthan objects in an STL container.

If your class needs to be copyable,prefer providing a copy method, such as CopyFrom() or Clone(), rather than acopy constructor, because such methods cannot be invoked implicitly. If a copymethod is insufficient in your situation (e.g. for performance reasons, orbecause your class needs to be stored by value in an STL container), provideboth a copy constructor and assignment operator.

If your class does not need a copyconstructor or assignment operator, you must explicitly disable them. To do so,add dummy declarations for the copy constructor and assignment operator in the private: section of your class, but do notprovide any corresponding definition (so that any attempt to use them resultsin a link error).

For convenience, a DISALLOW_COPY_AND_ASSIGN macro can be used:

// A macro to disallow the copy constructorand operator= functions

// This should be used in the private:declarations for a class

#define DISALLOW_COPY_AND_ASSIGN(TypeName)\

 TypeName(const TypeName&);               \

 void operator=(const TypeName&)

Then, in class Foo:

class Foo {

 public:

 Foo(int f);

 ~Foo();

 

 private:

 DISALLOW_COPY_AND_ASSIGN(Foo);

};

Structs vs. Classes

link▽

Use a struct only for passive objects that carry data;everything else is a class.

The struct and class keywordsbehave almost identically in C++. We add our own semantic meanings to eachkeyword, so you should use the appropriate keyword for the data-type you'redefining.

structs should be used for passive objects that carry data,and may have associated constants, but lack any functionality other thanaccess/setting the data members. The accessing/setting of fields is done bydirectly accessing the fields rather than through method invocations. Methodsshould not provide behavior but should only be used to set up the data members,e.g., constructor, destructor, Initialize(), Reset(), Validate().

If more functionality is required, a class is more appropriate. If in doubt,make it a class.

For consistency with STL, you can use struct instead of class for functors and traits.

Note that member variables in structsand classes have differentnaming rules.

Inheritance

link▽

Composition is often more appropriatethan inheritance. When using inheritance, make it public.

Definition:When a sub-class inherits from a baseclass, it includes the definitions of all the data and operations that theparent base class defines. In practice, inheritance is used in two major waysin C++: implementation inheritance, in which actual code is inherited by thechild, and interfaceinheritance, in which only method names are inherited.

Pros:Implementation inheritance reduces codesize by re-using the base class code as it specializes an existing type.Because inheritance is a compile-time declaration, you and the compiler canunderstand the operation and detect errors. Interface inheritance can be usedto programmatically enforce that a class expose a particular API. Again, thecompiler can detect errors, in this case, when a class does not define anecessary method of the API.

Cons:For implementation inheritance, becausethe code implementing a sub-class is spread between the base and the sub-class,it can be more difficult to understand an implementation. The sub-class cannotoverride functions that are not virtual, so the sub-class cannot change implementation.The base class may also define some data members, so that specifies physicallayout of the base class.

Decision:

All inheritance should be public. If you want to do private inheritance,you should be including an instance of the base class as a member instead.

Do not overuse implementationinheritance. Composition is often more appropriate. Try to restrict use ofinheritance to the "is-a" case: Bar subclasses Foo if it can reasonably be said that Bar "is a kind of" Foo.

Make your destructor virtual if necessary. If your class hasvirtual methods, its destructor should be virtual.

Limit the use of protected to those member functions thatmight need to be accessed from subclasses. Note that datamembers should be private.

When redefining an inherited virtualfunction, explicitly declare it virtual in thedeclaration of the derived class. Rationale: If virtual is omitted, the reader has to check all ancestorsof the class in question to determine if the function is virtual or not.

Multiple Inheritance

link▽

Only very rarely is multipleimplementation inheritance actually useful. We allow multiple inheritance onlywhen at most one of the base classes has an implementation; all other baseclasses must be pureinterface classes tagged with the Interface suffix.

Definition:Multiple inheritance allows a sub-classto have more than one base class. We distinguish between base classes that are pureinterfaces and those that have an implementation.

Pros:Multiple implementation inheritance maylet you re-use even more code than single inheritance (see Inheritance).

Cons:Only very rarely is multiple implementation inheritanceactually useful. When multiple implementation inheritance seems like thesolution, you can usually find a different, more explicit, and cleanersolution.

Decision:Multiple inheritance is allowed onlywhen all superclasses, with the possible exception of the first one, are pureinterfaces. In order to ensure that they remain pure interfaces, they mustend with the Interface suffix.

Note: There is an exception tothis rule on Windows.

Interfaces

link▽

Classes that satisfy certain conditionsare allowed, but not required, to end with an Interface suffix.

Definition:

A class is a pure interface if it meetsthe following requirements:

• It has only public pure virtual ("= 0") methods and static methods (but see below for destructor).
• It may not have non-static data members.
• It need not have any constructors defined. If a constructor is provided, it must take no arguments and it must be protected.
• If it is a subclass, it may only be derived from classes that satisfy these conditions and are tagged with the Interface suffix.

An interface class can never be directlyinstantiated because of the pure virtual method(s) it declares. To make sureall implementations of the interface can be destroyed correctly, the interfacemust also declare a virtual destructor (in an exception to the first rule, thisshould not be pure). See Stroustrup, The C++ Programming Language,3rd edition, section 12.4 for details.

Pros:Tagging a class with the Interface suffix lets others know that theymust not add implemented methods or non static data members. This isparticularly important in the case of multipleinheritance. Additionally, the interface concept is already well-understoodby Java programmers.

Cons:The Interface suffix lengthens the class name, which can make itharder to read and understand. Also, the interface property may be consideredan implementation detail that shouldn't be exposed to clients.

Decision:A class may end with Interface only if it meets the aboverequirements. We do not require the converse, however: classes that meet theabove requirements are not required to end with Interface.

Operator Overloading

link▽

Do not overload operators except inrare, special circumstances.

Definition:A class can define that operators suchas + and / operate on the class as if it were a built-in type.

Pros:Can make code appear more intuitivebecause a class will behave in the same way as built-in types (such as int). Overloaded operators are more playfulnames for functions that are less-colorfully named, such as Equals() or Add(). For some template functions to work correctly, you mayneed to define operators.

Cons:While operator overloading can make codemore intuitive, it has several drawbacks:

• It can fool our intuition into thinking that expensive operations are cheap, built-in operations.
• It is much harder to find the call sites for overloaded operators. Searching for Equals() is much easier than searching for relevant invocations of ==.
• Some operators work on pointers too, making it easy to introduce bugs. Foo + 4 may do one thing, while &Foo + 4 does something totally different. The compiler does not complain for either of these, making this very hard to debug.

Overloading also has surprising ramifications.For instance, if a class overloads unary operator&, it cannot safely be forward-declared.

Decision:

In general, do not overload operators.The assignment operator (operator=), in particular, is insidious andshould be avoided. You can define functions like Equals() andCopyFrom() if you need them. Likewise, avoidthe dangerous unary operator& at all costs, if there's anypossibility the class might be forward-declared.

However, there may be rare cases whereyou need to overload an operator to interoperate with templates or"standard" C++ classes (such as operator<<(ostream&, const T&) for logging). These are acceptableif fully justified, but you should try to avoid these whenever possible. Inparticular, do not overload operator== or operator< just so that your class can be used as a key in anSTL container; instead, you should create equality and comparison functor typeswhen declaring the container.

Some of the STL algorithms do requireyou to overload operator==, and you may do so in these cases,provided you document why.

See also CopyConstructors and FunctionOverloading.

Access Control

link▽

Make data members private, and provide access to them throughaccessor functions as needed (for technical reasons, we allow data members of atest fixture class to beprotected when using Google Test). Typically avariable would be called foo_ and the accessor function foo(). You may also want a mutator function set_foo(). Exception:static const data members (typically called kFoo) need not be private.

The definitions of accessors are usuallyinlined in the header file.

See also Inheritance and FunctionNames.

Declaration Order

link▽

Use the specified order of declarationswithin a class: public: before private:, methods before data members(variables), etc.

Your class definition should start withits public: section, followed by its protected: section and then its private: section. If any of these sectionsare empty, omit them.

Within each section, the declarationsgenerally should be in the following order:

• Typedefs and Enums
• Constants (static const data members)
• Constructors
• Destructor
• Methods, including static methods
• Data Members (except static const data members)

Friend declarations should always be inthe private section, and the DISALLOW_COPY_AND_ASSIGN macro invocation should be at the end of the private: section. It should be the lastthing in the class. See CopyConstructors.

Method definitions in the corresponding .cc file should be the same as thedeclaration order, as much as possible.

Do not put large method definitionsinline in the class definition. Usually, only trivial or performance-critical,and very short, methods may be defined inline. See InlineFunctions for more details.

Write Short Functions

link▽

Prefer small and focused functions.

We recognize that long functions aresometimes appropriate, so no hard limit is placed on functions length. If afunction exceeds about 40 lines, think about whether it can be broken upwithout harming the structure of the program.

Even if your long function worksperfectly now, someone modifying it in a few months may add new behavior. Thiscould result in bugs that are hard to find. Keeping your functions short andsimple makes it easier for other people to read and modify your code.

You could find long and complicatedfunctions when working with some code. Do not be intimidated by modifyingexisting code: if working with such a function proves to be difficult, you findthat errors are hard to debug, or you want to use a piece of it in severaldifferent contexts, consider breaking up the function into smaller and moremanageable pieces.

Google-Specific Magic

There are various tricks and utilitiesthat we use to make C++ code more robust, and various ways we use C++ that maydiffer from what you see elsewhere.

Smart Pointers

link▽

If you actually need pointer semantics, unique_ptr is great, and scoped_ptr is fine if you need to support olderversions of C++. You should only use shared_ptr with a non-const referent when it is trulynecessary to share ownership of an object (e.g. inside an STL container). Youshould never use auto_ptr.

Definition:"Smart" pointers are objectsthat act like pointers, but automate management of the underlying memory.

Pros:Smart pointers are extremely useful forpreventing memory leaks, and are essential for writing exception-safe code.They also formalize and document the ownership of dynamically allocated memory.

Cons:We prefer designs in which objects havesingle, fixed owners. Smart pointers which enable sharing or transfer ofownership can act as a tempting alternative to a careful design of ownershipsemantics, leading to confusing code and even bugs in which memory is neverdeleted. The semantics of smart pointers (especially auto_ptr) can be nonobvious and confusing. Theexception-safety benefits of smart pointers are not decisive, since we do notallow exceptions.

Decision:

unique_ptr

See below.

scoped_ptr

Prefer unique_ptr unless C++03 compatibility is required.

auto_ptr

Confusing and bug-prone ownership-transfer semantics. Use unique_ptr instead, if possible.

shared_ptr

Safe with const referents (i.e. shared_ptr<const T>). Reference-counted pointers withnon-const referents can occasionally be the best design, but try to rewritewith single owners where possible.

cpplint

link▽

Use cpplint.py to detect style errors.

cpplint.py is a tool that reads a source file and identifiesmany style errors. It is not perfect, and has both false positives and falsenegatives, but it is still a valuable tool. False positives can be ignored byputting // NOLINT at the end of the line.

Some projects have instructions on howto run cpplint.py from their project tools. If theproject you are contributing to does not, you can download cpplint.py separately.

Other C++ Features

Reference Arguments

link▽

All parameters passed by reference mustbe labeled const.

Definition:In C, if a function needs to modify avariable, the parameter must use a pointer, eg int foo(int *pval). In C++, the function can alternativelydeclare a reference parameter:intfoo(int &val).

Pros:Defining a parameter as reference avoidsugly code like (*pval)++. Necessary for some applications likecopy constructors. Makes it clear, unlike with pointers, that a null pointer isnot a possible value.

Cons:References can be confusing, as theyhave value syntax but pointer semantics.

Decision:

Within function parameter lists allreferences must be const:

void Foo(const string &in, string*out);

In fact it is a very strong conventionin Google code that input arguments are values or const references while output arguments are pointers.Input parameters may be constpointers, but we never allow non-const reference parameters except whenrequired by convention, e.g., swap().

However, there are some instances whereusing const T* is preferable to const T& for input parameters. For example:

• You want to pass in a null pointer.
• The function saves a pointer or reference to the input.

Remember that most of the time inputparameters are going to be specified as const T&. Using const T* insteadcommunicates to the reader that the input is somehow treated differently. So ifyou choose const T* rather than const T&, do so for a concrete reason; otherwiseit will likely confuse readers by making them look for an explanation thatdoesn't exist.

Function Overloading

link▽

Use overloaded functions (includingconstructors) only if a reader looking at a call site can get a good idea ofwhat is happening without having to first figure out exactly which overload isbeing called.

Definition:

You may write a function that takes a const string& and overload it with another thattakes const char*.

class MyClass {

 public:

 void Analyze(const string &text);

 void Analyze(const char *text, size_t textlen);

};

Pros:Overloading can make code more intuitiveby allowing an identically-named function to take different arguments. It maybe necessary for templatized code, and it can be convenient for Visitors.

Cons:If a function is overloaded by theargument types alone, a reader may have to understand C++'s complex matchingrules in order to tell what's going on. Also many people are confused by thesemantics of inheritance if a derived class overrides only some of the variantsof a function.

Decision:If you want to overload a function,consider qualifying the name with some information about the arguments, e.g., AppendString(), AppendInt() rather than just Append().

Default Arguments

link▽

We do not allow default functionparameters, except in limited situations as explained below. Simulate them withfunction overloading instead, if appropriate.

Pros:Often you have a function that usesdefault values, but occasionally you want to override the defaults. Defaultparameters allow an easy way to do this without having to define many functionsfor the rare exceptions. Compared to overloading the function, defaultarguments have a cleaner syntax, with less boilerplate and a clearerdistinction between 'required' and 'optional' arguments.

Cons:Function pointers are confusing in thepresence of default arguments, since the function signature often doesn't matchthe call signature. Adding a default argument to an existing function changesits type, which can cause problems with code taking its address. Addingfunction overloads avoids these problems. In addition, default parameters mayresult in bulkier code since they are replicated at every call-site -- asopposed to overloaded functions, where "the default" appears only inthe function definition.

Decision:

While the cons above are not thatonerous, they still outweigh the (small) benefits of default arguments overfunction overloading. So except as described below, we require all arguments tobe explicitly specified.

One specific exception is when thefunction is a static function (or in an unnamed namespace) in a .cc file. Inthis case, the cons don't apply since the function's use is so localized.

Another specific exception is whendefault arguments are used to simulate variable-length argument lists.

// Support up to 4 params by using adefault empty AlphaNum.

string StrCat(const AlphaNum &a,

              const AlphaNum &b =gEmptyAlphaNum,

              const AlphaNum &c =gEmptyAlphaNum,

              const AlphaNum &d =gEmptyAlphaNum);

Variable-Length Arrays and alloca()

link▽

We do not allow variable-length arraysor alloca().

Pros:Variable-length arrays havenatural-looking syntax. Both variable-length arrays and alloca() are very efficient.

Cons:Variable-length arrays and alloca arenot part of Standard C++. More importantly, they allocate a data-dependentamount of stack space that can trigger difficult-to-find memory overwritingbugs: "It ran fine on my machine, but dies mysteriously inproduction".

Decision:Use a safe allocator instead, such as scoped_ptr/scoped_array.

Friends

link▽

We allow use of friend classes and functions, withinreason.

Friends should usually be defined in thesame file so that the reader does not have to look in another file to find usesof the private members of a class. A common use offriend is to have a FooBuilder class be a friend of Foo so that it can construct the inner state of Foo correctly, without exposing thisstate to the world. In some cases it may be useful to make a unittest class afriend of the class it tests.

Friends extend, but do not break, theencapsulation boundary of a class. In some cases this is better than making amember public when you want to give only one other class access to it. However,most classes should interact with other classes solely through their publicmembers.

Exceptions

link▽

We do not use C++ exceptions.

Pros:

• Exceptions allow higher levels of an application to decide how to handle "can't happen" failures in deeply nested functions, without the obscuring and error-prone bookkeeping of error codes.
• Exceptions are used by most other modern languages. Using them in C++ would make it more consistent with Python, Java, and the C++ that others are familiar with.
• Some third-party C++ libraries use exceptions, and turning them off internally makes it harder to integrate with those libraries.
• Exceptions are the only way for a constructor to fail. We can simulate this with a factory function or an Init() method, but these require heap allocation or a new "invalid" state, respectively.
• Exceptions are really handy in testing frameworks.

Cons:

• When you add a throw statement to an existing function, you must examine all of its transitive callers. Either they must make at least the basic exception safety guarantee, or they must never catch the exception and be happy with the program terminating as a result. For instance, if f() calls g() calls h(), and h throws an exception that f catches, g has to be careful or it may not clean up properly.
• More generally, exceptions make the control flow of programs difficult to evaluate by looking at code: functions may return in places you don't expect. This causes maintainability and debugging difficulties. You can minimize this cost via some rules on how and where exceptions can be used, but at the cost of more that a developer needs to know and understand.
• Exception safety requires both RAII and different coding practices. Lots of supporting machinery is needed to make writing correct exception-safe code easy. Further, to avoid requiring readers to understand the entire call graph, exception-safe code must isolate logic that writes to persistent state into a "commit" phase. This will have both benefits and costs (perhaps where you're forced to obfuscate code to isolate the commit). Allowing exceptions would force us to always pay those costs even when they're not worth it.
• Turning on exceptions adds data to each binary produced, increasing compile time (probably slightly) and possibly increasing address space pressure.
• The availability of exceptions may encourage developers to throw them when they are not appropriate or recover from them when it's not safe to do so. For example, invalid user input should not cause exceptions to be thrown. We would need to make the style guide even longer to document these restrictions!

Decision:

On their face, the benefits of usingexceptions outweigh the costs, especially in new projects. However, forexisting code, the introduction of exceptions has implications on all dependentcode. If exceptions can be propagated beyond a new project, it also becomesproblematic to integrate the new project into existing exception-free code.Because most existing C++ code at Google is not prepared to deal withexceptions, it is comparatively difficult to adopt new code that generatesexceptions.

Given that Google's existing code is notexception-tolerant, the costs of using exceptions are somewhat greater than thecosts in a new project. The conversion process would be slow and error-prone.We don't believe that the available alternatives to exceptions, such as errorcodes and assertions, introduce a significant burden.

Our advice against using exceptions isnot predicated on philosophical or moral grounds, but practical ones. Becausewe'd like to use our open-source projects at Google and it's difficult to do soif those projects use exceptions, we need to advise against exceptions inGoogle open-source projects as well. Things would probably be different if wehad to do it all over again from scratch.

There is an exception tothis rule (no pun intended) for Windows code.

Run-Time Type Information (RTTI)

link▽

Avoid using Run Time Type Information(RTTI).

Definition:RTTI allows a programmer to query theC++ class of an object at run time. This is done by use of typeid or dynamic_cast.

Cons:

Querying the type of an object atrun-time frequently means a design problem. Needing to know the type of anobject at runtime is often an indication that the design of your classhierarchy is flawed.

Undisciplined use of RTTI makes codehard to maintain. It can lead to type-based decision trees or switch statementsscattered throughout the code, all of which must be examined when makingfurther changes.

Pros:

The standard alternatives to RTTI(described below) require modification or redesign of the class hierarchy inquestion. Sometimes such modifications are infeasible or undesirable,particularly in widely-used or mature code.

RTTI can be useful in some unit tests.For example, it is useful in tests of factory classes where the test has toverify that a newly created object has the expected dynamic type. It is alsouseful in managing the relationship between objects and their mocks.

RTTI is useful when considering multipleabstract objects. Consider

bool Base::Equal(Base* other) = 0;

bool Derived::Equal(Base* other) {

 Derived* that = dynamic_cast<Derived*>(other);

  if(that == NULL)

   return false;

  ...

}

Decision:

RTTI has legitimate uses but is prone toabuse, so you must be careful when using it. You may use it freely inunittests, but avoid it when possible in other code. In particular, think twicebefore using RTTI in new code. If you find yourself needing to write code thatbehaves differently based on the class of an object, consider one of thefollowing alternatives to querying the type:

• Virtual methods are the preferred way of executing different code paths depending on a specific subclass type. This puts the work within the object itself.
• If the work belongs outside the object and instead in some processing code, consider a double-dispatch solution, such as the Visitor design pattern. This allows a facility outside the object itself to determine the type of class using the built-in type system.

When the logic of a program guaranteesthat a given instance of a base class is in fact an instance of a particularderived class, then a dynamic_cast may be used freely on the object.Usually one can use a static_cast as an alternative in suchsituations.

Decision trees based on type are astrong indication that your code is on the wrong track.

if (typeid(*data) == typeid(D1)) {

  ...

} else if (typeid(*data) == typeid(D2)) {

  ...

} else if (typeid(*data) == typeid(D3)) {

...

Code such as this usually breaks whenadditional subclasses are added to the class hierarchy. Moreover, whenproperties of a subclass change, it is difficult to find and modify all theaffected code segments.

Do not hand-implement an RTTI-likeworkaround. The arguments against RTTI apply just as much to workarounds likeclass hierarchies with type tags. Moreover, workarounds disguise your trueintent.

Casting

link▽

Use C++ casts like static_cast<>(). Do not use other cast formats like int y = (int)x; or int y = int(x);.

Definition:C++ introduced a different cast systemfrom C that distinguishes the types of cast operations.

Pros:The problem with C casts is theambiguity of the operation; sometimes you are doing a conversion (e.g., (int)3.5) and sometimes you are doing a cast (e.g., (int)"hello"); C++ casts avoid this. AdditionallyC++ casts are more visible when searching for them.

Cons:The syntax is nasty.

Decision:

Do not use C-style casts. Instead, usethese C++-style casts.

• Use static_cast as the equivalent of a C-style cast that does value conversion, or when you need to explicitly up-cast a pointer from a class to its superclass.
• Use const_cast to remove the const qualifier (see const).
• Use reinterpret_cast to do unsafe conversions of pointer types to and from integer and other pointer types. Use this only if you know what you are doing and you understand the aliasing issues.

See the RTTIsection for guidance on the use of dynamic_cast.

Streams

link▽

Use streams only for logging.

Definition:Streams are a replacement for printf() and scanf().

Pros:With streams, you do not need to knowthe type of the object you are printing. You do not have problems with formatstrings not matching the argument list. (Though with gcc, you do not have thatproblem with printf either.) Streams have automaticconstructors and destructors that open and close the relevant files.

Cons:Streams make it difficult to dofunctionality like pread(). Some formatting (particularly thecommon format string idiom %.*s) is difficult if not impossible to doefficiently using streams without using printf-like hacks. Streams do not support operator reordering(the %1s directive), which is helpful forinternationalization.

Decision:

Do not use streams, except whererequired by a logging interface. Use printf-like routines instead.

There are various pros and cons to usingstreams, but in this case, as in many other cases, consistency trumps thedebate. Do not use streams in your code.

Extended Discussion

There has been debate on this issue, sothis explains the reasoning in greater depth. Recall the Only One Way guidingprinciple: we want to make sure that whenever we do a certain type of I/O, thecode looks the same in all those places. Because of this, we do not want toallow users to decide between using streams or using printf plus Read/Write/etc. Instead, weshould settle on one or the other. We made an exception for logging because itis a pretty specialized application, and for historical reasons.

Proponents of streams have argued thatstreams are the obvious choice of the two, but the issue is not actually soclear. For every advantage of streams they point out, there is an equivalentdisadvantage. The biggest advantage is that you do not need to know the type ofthe object to be printing. This is a fair point. But, there is a downside: youcan easily use the wrong type, and the compiler will not warn you. It is easyto make this kind of mistake without knowing when using streams.

cout << this;  // Prints the address

cout << *this;  // Prints the contents

The compiler does not generate an errorbecause << has been overloaded. We discourageoverloading for just this reason.

Some say printf formatting is ugly and hard to read, but streamsare often no better. Consider the following two fragments, both with the sametypo. Which is easier to discover?

cerr << "Error connecting to'" << foo->bar()->hostname.first

    << ":" << foo->bar()->hostname.second<< ": " << strerror(errno);

 

fprintf(stderr, "Error connecting to'%s:%u: %s",

       foo->bar()->hostname.first, foo->bar()->hostname.second,

       strerror(errno));

And so on and so forth for any issue youmight bring up. (You could argue, "Things would be better with the rightwrappers," but if it is true for one scheme, is it not also true for theother? Also, remember the goal is to make the language smaller, not add yetmore machinery that someone has to learn.)

Either path would yield differentadvantages and disadvantages, and there is not a clearly superior solution. Thesimplicity doctrine mandates we settle on one of them though, and the majoritydecision was on printf + read/write.

Preincrement and Predecrement

link▽

Use prefix form (++i) of the increment and decrementoperators with iterators and other template objects.

Definition:When a variable is incremented (++i or i++) or decremented (--i or i--) and the valueof the expression is not used, one must decide whether to preincrement(decrement) or postincrement (decrement).

Pros:When the return value is ignored, the"pre" form (++i) is never less efficient than the"post" form (i++), and is often more efficient. This isbecause post-increment (or decrement) requires a copy of i to be made, which is the value of the expression.If i is an iterator or other non-scalartype, copying i could be expensive. Since the twotypes of increment behave the same when the value is ignored, why not justalways pre-increment?

Cons:The tradition developed, in C, of usingpost-increment when the expression value is not used, especially in for loops. Some find post-incrementeasier to read, since the "subject" (i) precedes the "verb" (++), just like in English.

Decision:For simple scalar (non-object) valuesthere is no reason to prefer one form and we allow either. For iterators andother template types, use pre-increment.

Use of const

link▽

Use const whenever it makes sense.

Definition:Declared variables and parameters can bepreceded by the keyword const to indicate the variables are notchanged (e.g., const int foo). Class functions can have the constqualifier to indicate the function doesnot change the state of the class member variables (e.g., class Foo { int Bar(char c) const; };).

Pros:Easier for people to understand howvariables are being used. Allows the compiler to do better type checking, and,conceivably, generate better code. Helps people convince themselves of programcorrectness because they know the functions they call are limited in how theycan modify your variables. Helps people know what functions are safe to usewithout locks in multi-threaded programs.

Cons:const is viral: if you pass a const variable to a function, thatfunction must have const in its prototype (or the variablewill need a const_cast). This can be a particular problem whencalling library functions.

Decision:

const variables, data members, methods and arguments adda level of compile-time type checking; it is better to detect errors as soon aspossible. Therefore we strongly recommend that you use const whenever it makes sense to do so:

• If a function does not modify an argument passed by reference or by pointer, that argument should be const.
• Declare methods to be const whenever possible. Accessors should almost always be const. Other methods should be const if they do not modify any data members, do not call any non-const methods, and do not return a non-const pointer or non-const reference to a data member.
• Consider making data members const whenever they do not need to be modified after construction.

The mutable keyword is allowed but is unsafe when used withthreads, so thread safety should be carefully considered first.

Where to put the const

Some people favor the form int const *foo to const int* foo. They argue that this is more readable because it's moreconsistent: it keeps the rule that const always follows the object it's describing. However,this consistency argument doesn't apply in codebases with few deeply-nestedpointer expressions since most const expressionshave only one const, and it applies to the underlyingvalue. In such cases, there's no consistency to maintain. Putting the const first is arguably more readable,since it follows English in putting the "adjective" (const) before the "noun" (int).

That said, while we encourage putting const first, we do not require it. Butbe consistent with the code around you!

Integer Types

link▽

Of the built-in C++ integer types, theonly one used is int. If a program needs a variable of adifferent size, use a precise-width integer type from <stdint.h>, such asint16_t. If your variable represents a value that could ever begreater than or equal to 2^31 (2GiB), use a 64-bit type such as int64_t. Keep in mind that even if your valuewon't ever be too large for an int, it may be usedin intermediate calculations which may require a larger type. When in doubt,choose a larger type.

Definition:C++ does not specify the sizes of itsinteger types. Typically people assume that short is 16 bits, int is 32 bits, long is 32 bits and long long is 64 bits.

Pros:Uniformity of declaration.

Cons:The sizes of integral types in C++ canvary based on compiler and architecture.

Decision:

<stdint.h> defines types like int16_t, uint32_t, int64_t, etc. You should always use those in preference to short, unsigned long long and the like, when you need aguarantee on the size of an integer. Of the C integer types, only int should be used. When appropriate,you are welcome to use standard types like size_t and ptrdiff_t.

We use int very often, for integers we know are not going tobe too big, e.g., loop counters. Use plain old int for such things. You should assume that an int is at least 32 bits, but don'tassume that it has more than 32 bits. If you need a 64-bit integer type, use int64_t or uint64_t.

For integers we know can be"big", use int64_t.

You should not use the unsigned integertypes such as uint32_t, unless there is a valid reason such asrepresenting a bit pattern rather than a number, or you need defined overflowmodulo 2^N. In particular, do not use unsigned types to say a number will neverbe negative. Instead, use assertions for this.

If your code is a container that returnsa size, be sure to use a type that will accommodate any possible usage of yourcontainer. When in doubt, use a larger type rather than a smaller type.

Use care when converting integer types.Integer conversions and promotions can cause non-intuitive behavior.

On Unsigned Integers

Some people, including some textbookauthors, recommend using unsigned types to represent numbers that are nevernegative. This is intended as a form of self-documentation. However, in C, theadvantages of such documentation are outweighed by the real bugs it canintroduce. Consider:

for (unsigned int i = foo.Length()-1; i>= 0; --i) ...

This code will never terminate!Sometimes gcc will notice this bug and warn you, but often it will not. Equallybad bugs can occur when comparing signed and unsigned variables. Basically, C'stype-promotion scheme causes unsigned types to behave differently than onemight expect.

So, document that a variable isnon-negative using assertions. Don't use an unsigned type.

64-bit Portability

link▽

Code should be 64-bit and 32-bitfriendly. Bear in mind problems of printing, comparisons, and structurealignment.

• printf() specifiers for some types are not cleanly portable between 32-bit and 64-bit systems. C99 defines some portable format specifiers. Unfortunately, MSVC 7.1 does not understand some of these specifiers and the standard is missing a few, so we have to define our own ugly versions in some cases (in the style of the standard include file inttypes.h):

·        //printf macros for size_t, in the style of inttypes.h

·        #ifdef_LP64

·        #define__PRIS_PREFIX "z"

·        #else

·        #define__PRIS_PREFIX

·        #endif

·         

·        //Use these macros after a % in a printf format string

·        //to get correct 32/64 bit behavior, like this:

·        //size_t size = records.size();

·        //printf("%"PRIuS"\n", size);

·         

·        #definePRIdS __PRIS_PREFIX "d"

·        #definePRIxS __PRIS_PREFIX "x"

·        #definePRIuS __PRIS_PREFIX "u"

·        #definePRIXS __PRIS_PREFIX "X"

#define PRIoS __PRIS_PREFIX "o"

Type

DO NOT use

DO use

Notes

void * (or any pointer)

%lx

%p

int64_t

%qd, %lld

%"PRId64"

uint64_t

%qu, %llu, %llx

%"PRIu64", %"PRIx64"

size_t

%u

%"PRIuS", %"PRIxS"

C99 specifies %zu

ptrdiff_t

%d

%"PRIdS"

C99 specifies %zd

Note that the PRI* macros expand to independentstrings which are concatenated by the compiler. Hence if you are using anon-constant format string, you need to insert the value of the macro into theformat, rather than the name. It is still possible, as usual, to include lengthspecifiers, etc., after the % when using the PRI* macros. So, e.g. printf("x =%30"PRIuS"\n", x) would expand on 32-bit Linux to printf("x = %30" "u""\n", x), which the compiler will treat as printf("x = %30u\n", x).

• Remember that sizeof(void *) != sizeof(int). Use intptr_t if you want a pointer-sized integer.
• You may need to be careful with structure alignments, particularly for structures being stored on disk. Any class/structure with a int64_t/uint64_t member will by default end up being 8-byte aligned on a 64-bit system. If you have such structures being shared on disk between 32-bit and 64-bit code, you will need to ensure that they are packed the same on both architectures. Most compilers offer a way to alter structure alignment. For gcc, you can use __attribute__((packed)). MSVC offers#pragma pack() and __declspec(align()).
• Use the LL or ULL suffixes as needed to create 64-bit constants. For example:

·        int64_tmy_value = 0x123456789LL;

uint64_t my_mask = 3ULL << 48;

• If you really need different code on 32-bit and 64-bit systems, use #ifdef _LP64 to choose between the code variants. (But please avoid this if possible, and keep any such changes localized.)

Preprocessor Macros

link▽

Be very cautious with macros. Preferinline functions, enums, and const variablesto macros.

Macros mean that the code you see is notthe same as the code the compiler sees. This can introduce unexpected behavior,especially since macros have global scope.

Luckily, macros are not nearly asnecessary in C++ as they are in C. Instead of using a macro to inlineperformance-critical code, use an inline function. Instead of using a macro tostore a constant, use a const variable. Instead of using a macroto "abbreviate" a long variable name, use a reference. Instead ofusing a macro to conditionally compile code ... well, don't do that at all(except, of course, for the #define guards to prevent double inclusionof header files). It makes testing much more difficult.

Macros can do things these othertechniques cannot, and you do see them in the codebase, especially in thelower-level libraries. And some of their special features (like stringifying,concatenation, and so forth) are not available through the language proper. Butbefore using a macro, consider carefully whether there's a non-macro way toachieve the same result.

The following usage pattern will avoidmany problems with macros; if you use macros, follow it whenever possible:

• Don't define macros in a .h file.
• #define macros right before you use them, and #undef them right after.
• Do not just #undef an existing macro before replacing it with your own; instead, pick a name that's likely to be unique.
• Try not to use macros that expand to unbalanced C++ constructs, or at least document that behavior well.
• Prefer not using ## to generate function/class/variable names.

0 and nullptr/NULL

link▽

Use 0 for integers, 0.0 for reals, nullptr (or NULL) for pointers,and '\0' for chars.

Use 0 for integers and 0.0 for reals. This is not controversial.

For pointers (address values), there isa choice between 0 and NULL (and, for C++11, nullptr). For projects that allow C++11 features, use nullptr. For C++03 projects, we prefer NULL because it looks like a pointer.In fact, some C++ compilers provide special definitions of NULL which enable them to give usefulwarnings, particularly in situations where sizeof(NULL) is not equal to sizeof(0).

Use '\0' for chars. This is the correct type and also makescode more readable.

sizeof

link▽

Prefer sizeof(varname) to sizeof(type).

Use sizeof(varname) when you take the size of aparticular variable. sizeof(varname) will update appropriately ifsomeone changes the variable type either now or later. You may use sizeof(type) for code unrelated to anyparticular variable, such as code that manages an external or internal dataformat where a variable of an appropriate C++ type is not convenient.

Struct data;

memset(&data, 0, sizeof(data));

memset(&data, 0, sizeof(Struct));

if (raw_size < sizeof(int)) {

 LOG(ERROR) << "compressed record not big enough for count:" << raw_size;

 return false;

}

auto

link▽

Use auto to avoid type names that are just clutter. Continueto use manifest type declarations when it helps readability, and never use auto for anything but local variables.

Definition:In C++11, a variable whose type is givenas auto will be given a type that matchesthat of the expression used to initialize it. You can use auto either to initialize a variable bycopying, or to bind a reference.

vector<string> v;

...

auto s1 = v[0];  // Makes a copy of v[0].

const auto& s2 = v[0];  // s2 is a reference to v[0].

Pros:

C++ type names can sometimes be long andcumbersome, especially when they involve templates or namespaces. In astatement like

sparse_hash_map<string,int>::iterator iter = m.find(val);

the return type is hard to read, andobscures the primary purpose of the statement. Changing it to

auto iter = m.find(val);

makes it more readable.

Without auto we are sometimes forced to write a type name twicein the same expression, adding no value for the reader, as in

diagnostics::ErrorStatus* status = newdiagnostics::ErrorStatus("xyz");

Using auto makes it easier to use intermediate variables whenappropriate, by reducing the burden of writing their types explicitly.

Cons:

Sometimes code is clearer when types aremanifest, especially when a variable's initialization depends on things thatwere declared far away. In an expression like

auto i = x.Lookup(key);

it may not be obvious what i's type is, if x was declared hundreds of lines earlier.

Programmers have to understand thedifference between auto and const auto& or they'll get copies when they didn't mean to.

The interaction between auto and C++11 brace-initialization canbe confusing. The declarations

auto x(3); // Note: parentheses.

auto y{3}; // Note: curly braces.

mean different things — x is an int, while y is an initializer_list. The same applies to othernormally-invisible proxy types.

If an auto variable is used as part of an interface, e.g. as aconstant in a header, then a programmer might change its type while onlyintending to change its value, leading to a more radical API change thanintended.

Decision:

auto is permitted, for local variables only. Do not use auto for file-scope or namespace-scopevariables, or for class members. Never assign a braced initializer list to anauto-typed variable.

The auto keyword is also used in an unrelated C++11 feature:it's part of the syntax for a new kind of function declaration with a trailingreturn type. Function declarations with trailing return types are notpermitted.

Brace Initialization

link▽

You may use brace initialization.

In C++03, aggregate types (arrays andstructs with no constructor) could be initialized using braces.

struct Point { int x; int y; };

Point p = {1, 2};

In C++11, this syntax has been expandedfor use with all other datatypes. The brace initialization form is called braced-init-list.Here are a few examples of its use.

// Vector takes lists of elements.

vector<string> v{"foo","bar"};

 

// The same, except this form cannot beused if the initializer_list

// constructor is explicit. You may chooseto use either form.

vector<string> v = {"foo","bar"};

 

// Maps take lists of pairs. Nestedbraced-init-lists work.

map<int, string> m = {{1,"one"}, {2, "2"}};

 

// braced-init-lists can be implicitlyconverted to return types.

vector<int> test_function() {

 return {1, 2, 3};

}

 

// Iterate over a braced-init-list.

for (int i : {-1, -2, -3}) {}

 

// Call a function using abraced-init-list.

void test_function2(vector<int> v) {}

test_function2({1, 2, 3});

User data types can also defineconstructors that take initializer_list, which is automatically created from braced-init-list:

class MyType {

 public:

  //initializer_list is a reference to the underlying init list,

  // so it can be passed by value.

 MyType(initializer_list<int> init_list) {

   for (int element : init_list) {}

  }

};

MyType m{2, 3, 5, 7};

Finally, brace initialization can alsocall ordinary constructors of data types that do not have initializer_list constructors.

double d{1.23};

// Calls ordinary constructor as long asMyOtherType has no

// initializer_list constructor.

class MyOtherType {

 public:

 explicit MyOtherType(string);

 MyOtherType(int, string);

};

MyOtherType m = {1, "b"};

// If the constructor is explicit, youcan't use the "= {}" form.

MyOtherType m{"b"};

Never assign a braced-init-list toan auto local variable. In the single element case, what this means can beconfusing.

auto d = {1.23};        // d is an initializer_list<double>

auto d = double{1.23};  // Good -- d is a double, not aninitializer_list.

Boost

link▽

Use only approved libraries from theBoost library collection.

Definition:The Boostlibrary collection is a popular collection of peer-reviewed, free,open-source C++ libraries.

Pros:Boost code is generally veryhigh-quality, is widely portable, and fills many important gaps in the C++standard library, such as type traits, better binders, and better smartpointers. It also provides an implementation of the TR1 extension to the standardlibrary.

Cons:Some Boost libraries encourage codingpractices which can hamper readability, such as metaprogramming and otheradvanced template techniques, and an excessively "functional" styleof programming.

Decision:

In order to maintain a high level ofreadability for all contributors who might read and maintain code, we onlyallow an approved subset of Boost features. Currently, the following librariesare permitted:

• Call Traits from boost/call_traits.hpp
• Compressed Pair from boost/compressed_pair.hpp
• The Boost Graph Library (BGL) from boost/graph, except serialization (adj_list_serialize.hpp) and parallel/distributed algorithms and data structures (boost/graph/parallel/* and boost/graph/distributed/*).
• Property Map from boost/property_map, except parallel/distributed property maps (boost/property_map/parallel/*).
• The part of Iterator that deals with defining iterators: boost/iterator/iterator_adaptor.hpp, boost/iterator/iterator_facade.hpp, and boost/function_output_iterator.hpp
• The part of Polygon that deals with Voronoi diagram construction and doesn't depend on the rest of Polygon: boost/polygon/voronoi_builder.hpp,boost/polygon/voronoi_diagram.hpp, and boost/polygon/voronoi_geometry_type.hpp
• Bimap from boost/bimap

We are actively considering adding otherBoost features to the list, so this list may be expanded in the future.

The following libraries are permitted,but their use is discouraged because they've been superseded by standardlibraries in C++11:

• Array from boost/array.hpp: use std::array instead.
• Pointer Container from boost/ptr_container: use containers of std::unique_ptr instead.

C++11

link▽

Use only approved libraries and languageextensions from C++11 (formerly known as C++0x). Consider portability to otherenvironments before using C++11 features in your project.

Definition:C++11 is the latest ISO C++ standard. Itcontains significantchanges both to the language and libraries.

Pros:C++11 has become the official standard,and eventually will be supported by most C++ compilers. It standardizes somecommon C++ extensions that we use already, allows shorthands for someoperations, and has some performance and safety improvements.

Cons:

The C++11 standard is substantially morecomplex than its predecessor (1,300 pages versus 800 pages), and is unfamiliarto many developers. The long-term effects of some features on code readabilityand maintenance are unknown. We cannot predict when its various features willbe implemented uniformly by tools that may be of interest (gcc, icc, clang,Eclipse, etc.).

As with Boost,some C++11 extensions encourage coding practices that hamper readability—forexample by removing checked redundancy (such as type names) that may be helpfulto readers, or by encouraging template metaprogramming. Other extensionsduplicate functionality available through existing mechanisms, which may leadto confusion and conversion costs.

Decision:Use only C++11 libraries and languagefeatures that have been approved for use. Currently only the following C++11features are approved:

• auto (for local variables only).
• constexpr (for ensuring constants).
• Use of >> with no intervening space to close multiple levels of template arguments, as in set<list<string>>, where C++03 required a space as in set<list<string> >.
• Range-based for loops.
• Use of the LL and ULL suffixes on numeric literals to guarantee that their type is at least 64 bits wide.
• Variadic macros (but note that use of macros is discouraged).
• All of the new STL algorithms in the <algorithm> and <numeric> headers.
• Use of local types as template parameters.
• nullptr and nullptr_t.
• static_assert.
• Everything in <array>.
• Everything in <tuple>.
• Variadic templates.
• Alias templates (the new using syntax) may be used. Don't use an alias declaration where a typedef will work.
• unique_ptr, with restrictions (see below).
• Brace initialization syntax. See the full section for more detail.

Other features will be approvedindividually as appropriate. Avoid writing code that is incompatible with C++11(even though it works in C++03).

unique_ptr

link▽

Use unique_ptr freely within classes and functions, but do not useit to transfer ownership outside of a single .cc/.h pair.

Definition:unique_ptr is a "smart" pointer type which expressesexclusive ownership of the underlying object. It provides "movesemantics", which enables it to be stored in containers and used as afunction parameter or return value, even though it cannot be copied (topreserve the unique-ownership property).

Pros:

• It fully automates memory management of singly-owned pointers, virtually eliminating the risk of leaking the underlying memory.
• It provides a single, universal abstraction for singly-owned pointers, replacing close to a dozen overlapping partial solutions in common use at Google.
• It can be passed into and returned from functions, providing a self-documenting and compiler-enforced way to transfer ownership between scopes.
• Its performance is essentially identical to a plain pointer.

Cons:

• It cannot be used in code that requires C++03 compatibility.
• Move semantics implicitly rely on rvalue references, a new C++11 feature which is unfamiliar to many Googlers.
• Google code currently uses a completely different set of conventions for ownership transfer. Mixing unique_ptr with the existing conventions could add complexity and create confusion.
• Best practices for using unique_ptr within Google have not yet been established.

Decision:

Use of unique_ptr as a class member or local variable is encouraged,as is storing unique_ptr in containers that support it.However, for the time being it is forbidden to use unique_ptr as a function parameter or returnvalue, except for functions that are local to a single .cc/.h pair.

Note that the std::move() function, which is often used topass unique_ptr into function calls, remainsforbidden.

Naming

The most important consistency rules arethose that govern naming. The style of a name immediately informs us what sortof thing the named entity is: a type, a variable, a function, a constant, amacro, etc., without requiring us to search for the declaration of that entity.The pattern-matching engine in our brains relies a great deal on these namingrules.

Naming rules are pretty arbitrary, butwe feel that consistency is more important than individual preferences in thisarea, so regardless of whether you find them sensible or not, the rules are therules.

General Naming Rules

link▽

Function names, variable names, andfilenames should be descriptive; eschew abbreviation.

Give as descriptive a name as possible,within reason. Do not worry about saving horizontal space as it is far moreimportant to make your code immediately understandable by a new reader. Do notuse abbreviations that are ambiguous or unfamiliar to readers outside yourproject, and do not abbreviate by deleting letters within a word.

int price_count_reader;    // No abbreviation.

int num_errors;            // "num" is a widespreadconvention.

int num_dns_connections;   // Most people know what "DNS"stands for.

int n;                     // Meaningless.

int nerr;                  // Ambiguous abbreviation.

int n_comp_conns;          // Ambiguous abbreviation.

int wgc_connections;       // Only your group knows what thisstands for.

int pc_reader;             // Lots of things can beabbreviated "pc".

int cstmr_id;              // Deletes internal letters.

File Names

link▽

Filenames should be all lowercase andcan include underscores (_) or dashes (-). Follow the convention that your project uses. If thereis no consistent local pattern to follow, prefer "_".

Examples of acceptable file names:

my_useful_class.cc
my-useful-class.cc
myusefulclass.cc
myusefulclass_test.cc // _unittest and _regtest are deprecated.

C++ files should end in .cc and header files should end in .h.

Do not use filenames that already existin /usr/include, such as db.h.

In general, make your filenames veryspecific. For example, use http_server_logs.h rather than logs.h. A very common case is to have a pairof files called, e.g., foo_bar.hand foo_bar.cc, defining a class called FooBar.

Inline functions must be in a .h file. If your inline functions are very short, theyshould go directly into your .h file.However, if your inline functions include a lot of code, they may go into athird file that ends in -inl.h. In a class with a lot of inline code,your class could have three files:

url_table.h      // The class declaration.

url_table.cc     // The class definition.

url_table-inl.h  // Inline functions that include lots ofcode.

See also the section -inl.hFiles

Type Names

link▽

Type names start with a capital letterand have a capital letter for each new word, with no underscores: MyExcitingClass, MyExcitingEnum.

The names of all types — classes,structs, typedefs, and enums — have the same naming convention. Type namesshould start with a capital letter and have a capital letter for each new word.No underscores. For example:

// classes and structs

class UrlTable { ...

class UrlTableTester { ...

struct UrlTableProperties { ...

 

// typedefs

typedef hash_map<UrlTableProperties *, string>PropertiesMap;

 

// enums

enum UrlTableErrors { ...

Variable Names

link▽

Variable names are all lowercase, withunderscores between words. Class member variables have trailing underscores.For instance: my_exciting_local_variable,my_exciting_member_variable_.

Common Variable names

For example:

string table_name;  // OK - uses underscore.

string tablename;   // OK - all lowercase.

string tableName;   // Bad - mixed case.

Class Data Members

Data members (also called instancevariables or member variables) are lowercase with optional underscores likeregular variable names, but always end with a trailing underscore.

string table_name_;  // OK - underscore at end.

string tablename_;   // OK.

Struct Variables

Data members in structs should be namedlike regular variables without the trailing underscores that data members inclasses have.

struct UrlTableProperties {

 string name;

  intnum_entries;

}

See Structsvs. Classes for a discussion of when to use a struct versus a class.

Global Variables

There are no special requirements for globalvariables, which should be rare in any case, but if you use one, considerprefixing it with g_ or some other marker to easilydistinguish it from local variables.

Constant Names

link▽

Use a k followed by mixed case: kDaysInAWeek.

All compile-time constants, whether theyare declared locally, globally, or as part of a class, follow a slightlydifferent naming convention from other variables. Use a kfollowed by words with uppercase first letters:

const int kDaysInAWeek = 7;

Function Names

link▽

Regular functions have mixed case;accessors and mutators match the name of the variable: MyExcitingFunction(), MyExcitingMethod(), my_exciting_member_variable(),set_my_exciting_member_variable().

Regular Functions

Functions should start with a capitalletter and have a capital letter for each new word. No underscores.

If your function crashes upon an error,you should append OrDie to the function name. This only applies to functionswhich could be used by production code and to errors that are reasonably likelyto occur during normal operation.

AddTableEntry()

DeleteUrl()

OpenFileOrDie()

Accessors and Mutators

Accessors and mutators (get and setfunctions) should match the name of the variable they are getting and setting.This shows an excerpt of a class whose instance variable isnum_entries_.

class MyClass {

 public:

  ...

  intnum_entries() const { return num_entries_; }

 void set_num_entries(int num_entries) { num_entries_ = num_entries; }

 

 private:

  intnum_entries_;

};

You may also use lowercase letters forother very short inlined functions. For example if a function were so cheap youwould not cache the value if you were calling it in a loop, then lowercasenaming would be acceptable.

Namespace Names

link▽

Namespace names are all lower-case, andbased on project names and possibly their directory structure: google_awesome_project.

See Namespaces fora discussion of namespaces and how to name them.

Enumerator Names

link▽

Enumerators should be named either like constants orlike macros:either kEnumName or ENUM_NAME.

Preferably, the individual enumeratorsshould be named like constants.However, it is also acceptable to name them like macros.The enumeration name, UrlTableErrors (andAlternateUrlTableErrors), is a type, and therefore mixed case.

enum UrlTableErrors {

  kOK= 0,

 kErrorOutOfMemory,

 kErrorMalformedInput,

};

enum AlternateUrlTableErrors {

  OK= 0,

 OUT_OF_MEMORY = 1,

 MALFORMED_INPUT = 2,

};

Until January 2009, the style was toname enum values like macros.This caused problems with name collisions between enum values and macros.Hence, the change to prefer constant-style naming was put in place. New codeshould prefer constant-style naming if possible. However, there is no reason tochange old code to use constant-style names, unless the old names are actuallycausing a compile-time problem.

Macro Names

link▽

You're not really going to definea macro, are you? If you do, they're like this: MY_MACRO_THAT_SCARES_SMALL_CHILDREN.

Please see the descriptionof macros; in general macros should not be used. However,if they are absolutely needed, then they should be named with all capitals andunderscores.

#define ROUND(x) ...

#define PI_ROUNDED 3.0

Exceptions to Naming Rules

link▽

If you are naming something that isanalogous to an existing C or C++ entity then you can follow the existingnaming convention scheme.

bigopen()

function name, follows form of open()

uint

typedef

bigpos

struct or class, follows form of pos

sparse_hash_map

STL-like entity; follows STL naming conventions

LONGLONG_MAX

a constant, as in INT_MAX

Comments

Though a pain to write, comments areabsolutely vital to keeping our code readable. The following rules describewhat you should comment and where. But remember: while comments are veryimportant, the best code is self-documenting. Giving sensible names to typesand variables is much better than using obscure names that you must thenexplain through comments.

When writing your comments, write foryour audience: the next contributor who will need to understand your code. Begenerous — the next one may be you!

Comment Style

link▽

Use either the // or /* */ syntax, aslong as you are consistent.

You can use either the // or the /* */ syntax; however, // is much more common. Be consistentwith how you comment and what style you use where.

File Comments

link▽

Start each file with licenseboilerplate, followed by a description of its contents.

Legal Notice and Author Line

Every file should contain licenseboilerplate. Choose the appropriate boilerplate for the license used by theproject (for example, Apache 2.0, BSD, LGPL, GPL).

If you make significant changes to afile with an author line, consider deleting the author line.

File Contents

Every file should have a comment at thetop describing its contents.

Generally a .h file will describe the classes that are declared inthe file with an overview of what they are for and how they are used. A .cc file should contain moreinformation about implementation details or discussions of tricky algorithms.If you feel the implementation details or a discussion of the algorithms wouldbe useful for someone reading the .h, feel free to put it there instead, but mention in the .cc that the documentation is in the .h file.

Do not duplicate comments in both the .h and the .cc. Duplicated comments diverge.

Class Comments

link▽

Every class definition should have anaccompanying comment that describes what it is for and how it should be used.

// Iterates over the contents of aGargantuanTable.  Sample usage:

//   GargantuanTableIterator* iter = table->NewIterator();

//   for (iter->Seek("foo"); !iter->done(); iter->Next()){

//     process(iter->key(), iter->value());

//   }

//   delete iter;

class GargantuanTableIterator {

  ...

};

If you have already described a class indetail in the comments at the top of your file feel free to simply state"See comment at top of file for a complete description", but be sureto have some sort of comment.

Document the synchronization assumptionsthe class makes, if any. If an instance of the class can be accessed bymultiple threads, take extra care to document the rules and invariantssurrounding multithreaded use.

Function Comments

link▽

Declaration comments describe use of thefunction; comments at the definition of a function describe operation.

Function Declarations

Every function declaration should havecomments immediately preceding it that describe what the function does and howto use it. These comments should be descriptive ("Opens the file")rather than imperative ("Open the file"); the comment describes thefunction, it does not tell the function what to do. In general, these commentsdo not describe how the function performs its task. Instead, that should beleft to comments in the function definition.

Types of things to mention in commentsat the function declaration:

• What the inputs and outputs are.
• For class member functions: whether the object remembers reference arguments beyond the duration of the method call, and whether it will free them or not.
• If the function allocates memory that the caller must free.
• Whether any of the arguments can be a null pointer.
• If there are any performance implications of how a function is used.
• If the function is re-entrant. What are its synchronization assumptions?

Here is an example:

// Returns an iterator for this table.  It is the client's

// responsibility to delete the iteratorwhen it is done with it,

// and it must not use the iterator oncethe GargantuanTable object

// on which the iterator was created hasbeen deleted.

//

// The iterator is initially positioned atthe beginning of the table.

//

// This method is equivalent to:

//   Iterator* iter = table->NewIterator();

//   iter->Seek("");

//   return iter;

// If you are going to immediately seek toanother place in the

// returned iterator, it will be faster touse NewIterator()

// and avoid the extra seek.

Iterator* GetIterator() const;

However, do not be unnecessarily verboseor state the completely obvious. Notice below that it is not necessary to say"returns false otherwise" because this is implied.

// Returns true if the table cannot holdany more entries.

bool IsTableFull();

When commenting constructors anddestructors, remember that the person reading your code knows what constructorsand destructors are for, so comments that just say something like "destroysthis object" are not useful. Document what constructors do with theirarguments (for example, if they take ownership of pointers), and what cleanupthe destructor does. If this is trivial, just skip the comment. It is quitecommon for destructors not to have a header comment.

Function Definitions

If there is anything tricky about how afunction does its job, the function definition should have an explanatorycomment. For example, in the definition comment you might describe any codingtricks you use, give an overview of the steps you go through, or explain whyyou chose to implement the function in the way you did rather than using aviable alternative. For instance, you might mention why it must acquire a lockfor the first half of the function but why it is not needed for the secondhalf.

Note you should not justrepeat the comments given with the function declaration, in the .h file or wherever. It's okay to recapitulate brieflywhat the function does, but the focus of the comments should be on how it doesit.

Variable Comments

link▽

In general the actual name of thevariable should be descriptive enough to give a good idea of what the variableis used for. In certain cases, more comments are required.

Class Data Members

Each class data member (also called aninstance variable or member variable) should have a comment describing what itis used for. If the variable can take sentinel values with special meanings,such as a null pointer or -1, document this. For example:

private:

 //Keeps track of the total number of entries in the table.

 //Used to ensure we do not go over the limit. -1 means

 //that we don't yet know how many entries the table has.

 intnum_total_entries_;

Global Variables

As with data members, all globalvariables should have a comment describing what they are and what they are usedfor. For example:

// The total number of tests cases that werun through in this regression test.

const int kNumTestCases = 6;

Implementation Comments

link▽

In your implementation you should havecomments in tricky, non-obvious, interesting, or important parts of your code.

Class Data Members

Tricky or complicated code blocks shouldhave comments before them. Example:

// Divide result by two, taking intoaccount that x

// contains the carry from the add.

for (int i = 0; i < result->size();i++) {

  x =(x << 8) + (*result)[i];

 (*result)[i] = x >> 1;

  x&= 1;

}

Line Comments

Also, lines that are non-obvious shouldget a comment at the end of the line. These end-of-line comments should beseparated from the code by 2 spaces. Example:

// If we have enough memory, mmap the dataportion too.

mmap_budget = max<int64>(0,mmap_budget - index_->length());

if (mmap_budget >= data_size_ &&!MmapData(mmap_chunk_bytes, mlock))

 return;  // Error already logged.

Note that there are both comments thatdescribe what the code is doing, and comments that mention that an error hasalready been logged when the function returns.

If you have several comments onsubsequent lines, it can often be more readable to line them up:

DoSomething();                  // Comment here so thecomments line up.

DoSomethingElseThatIsLonger();  // Comment here so there are two spacesbetween

                                // the code andthe comment.

{ // One space before comment when openinga new scope is allowed,

  //thus the comment lines up with the following comments and code.

 DoSomethingElse();  // Two spacesbefore line comments normally.

}

DoSomething(); /* For trailing blockcomments, one space is fine. */

nullptr/NULL, true/false, 1, 2, 3...

When you pass in a null pointer,boolean, or literal integer values to functions, you should consider adding acomment about what they are, or make your code self-documenting by usingconstants. For example, compare:

bool success =CalculateSomething(interesting_value,

                                  10,

                                  false,

                                  NULL);  // What are these arguments??

versus:

bool success =CalculateSomething(interesting_value,

                                  10,     // Default base value.

                                  false,  // Not the first time we're calling this.

                                  NULL);  // No callback.

Or alternatively, constants orself-describing variables:

const int kDefaultBaseValue = 10;

const bool kFirstTimeCalling = false;

Callback *null_callback = NULL;

bool success =CalculateSomething(interesting_value,

                                 kDefaultBaseValue,

                                  kFirstTimeCalling,

                                 null_callback);

Don'ts

Note that you should never describethe code itself. Assume that the person reading the code knows C++ better thanyou do, even though he or she does not know what you are trying to do:

// Now go through the b array and make surethat if i occurs,

// the next element is i+1.

...       // Geez.  What a useless comment.

Punctuation, Spelling and Grammar

link▽

Pay attention to punctuation, spelling,and grammar; it is easier to read well-written comments than badly writtenones.

Comments should be as readable asnarrative text, with proper capitalization and punctuation. In many cases,complete sentences are more readable than sentence fragments. Shorter comments,such as comments at the end of a line of code, can sometimes be less formal,but you should be consistent with your style.

Although it can be frustrating to have acode reviewer point out that you are using a comma when you should be using asemicolon, it is very important that source code maintain a high level ofclarity and readability. Proper punctuation, spelling, and grammar help withthat goal.

TODO Comments

link▽

Use TODO comments for code that is temporary, a short-termsolution, or good-enough but not perfect.

TODOs should include the string TODO in all caps, followed by the name, e-mail address,or other identifier of the person who can best provide context about theproblem referenced by the TODO. A colon is optional. The main purposeis to have a consistent TODO format that can be searched tofind the person who can provide more details upon request. A TODO is not a commitment that theperson referenced will fix the problem. Thus when you create a TODO, it is almost always your name that isgiven.

// TODO([e-mail]): Use a "*"here for concatenation operator.

// TODO(Zeke) change this to use relations.

If your TODO is of the form "At a future date dosomething" make sure that you either include a very specific date("Fix by November 2005") or a very specific event ("Remove thiscode when all clients can handle XML responses.").

Deprecation Comments

link▽

Mark deprecated interface points with DEPRECATED comments.

You can mark an interface as deprecatedby writing a comment containing the word DEPRECATED in all caps. The comment goes either before thedeclaration of the interface or on the same line as the declaration.

After the word DEPRECATED, write your name, e-mail address, orother identifier in parentheses.

A deprecation comment must includesimple, clear directions for people to fix their callsites. In C++, you canimplement a deprecated function as an inline function that calls the newinterface point.

Marking an interface point DEPRECATED will not magically cause anycallsites to change. If you want people to actually stop using the deprecatedfacility, you will have to fix the callsites yourself or recruit a crew to helpyou.

New code should not contain calls todeprecated interface points. Use the new interface point instead. If you cannotunderstand the directions, find the person who created the deprecation and askthem for help using the new interface point.

Formatting

Coding style and formatting are prettyarbitrary, but a project is much easier to follow if everyone uses the samestyle. Individuals may not agree with every aspect of the formatting rules, andsome of the rules may take some getting used to, but it is important that allproject contributors follow the style rules so that they can all read andunderstand everyone's code easily.

To help you format code correctly, we'vecreated a settingsfile for emacs.

Line Length

link▽

Each line of text in your code should beat most 80 characters long.

We recognize that this rule iscontroversial, but so much existing code already adheres to it, and we feelthat consistency is important.

Pros:Those who favor this rule argue that itis rude to force them to resize their windows and there is no need for anythinglonger. Some folks are used to having several code windows side-by-side, andthus don't have room to widen their windows in any case. People set up theirwork environment assuming a particular maximum window width, and 80 columns hasbeen the traditional standard. Why change it?

Cons:Proponents of change argue that a widerline can make code more readable. The 80-column limit is an hidebound throwbackto 1960s mainframes; modern equipment has wide screens that can easily showlonger lines.

Decision:

80 characters is the maximum.

Exception: if a comment line contains anexample command or a literal URL longer than 80 characters, that line may belonger than 80 characters for ease of cut and paste.

Exception: an #include statement with a long path may exceed80 columns. Try to avoid situations where this becomes necessary.

Exception: you needn't be concernedabout headerguards that exceed the maximum length.

Non-ASCII Characters

link▽

Non-ASCII characters should be rare, andmust use UTF-8 formatting.

You shouldn't hard-code user-facing textin source, even English, so use of non-ASCII characters should be rare.However, in certain cases it is appropriate to include such words in your code.For example, if your code parses data files from foreign sources, it may beappropriate to hard-code the non-ASCII string(s) used in those data files asdelimiters. More commonly, unittest code (which does not need to be localized)might contain non-ASCII strings. In such cases, you should use UTF-8, sincethat is an encoding understood by most tools able to handle more than justASCII. Hex encoding is also OK, and encouraged where it enhances readability —for example, "\xEF\xBB\xBF" is the Unicode zero-width no-breakspace character, which would be invisible if included in the source as straightUTF-8.

Spaces vs. Tabs

link▽

Use only spaces, and indent 2 spaces ata time.

We use spaces for indentation. Do notuse tabs in your code. You should set your editor to emit spaces when you hitthe tab key.

Function Declarations and Definitions

link▽

Return type on the same line as functionname, parameters on the same line if they fit.

Functions look like this:

ReturnType ClassName::FunctionName(Typepar_name1, Type par_name2) {

 DoSomething();

  ...

}

If you have too much text to fit on oneline:

ReturnTypeClassName::ReallyLongFunctionName(Type par_name1, Type par_name2,

                                            Type par_name3) {

 DoSomething();

  ...

}

or if you cannot fit even the firstparameter:

ReturnTypeLongClassName::ReallyReallyReallyLongFunctionName(

   Type par_name1,  // 4 space indent

   Type par_name2,

   Type par_name3) {

 DoSomething();  // 2 space indent

  ...

}

Some points to note:

• The return type is always on the same line as the function name.
• The open parenthesis is always on the same line as the function name.
• There is never a space between the function name and the open parenthesis.
• There is never a space between the parentheses and the parameters.
• The open curly brace is always at the end of the same line as the last parameter.
• The close curly brace is either on the last line by itself or (if other style rules permit) on the same line as the open curly brace.
• There should be a space between the close parenthesis and the open curly brace.
• All parameters should be named, with identical names in the declaration and implementation.
• All parameters should be aligned if possible.
• Default indentation is 2 spaces.
• Wrapped parameters have a 4 space indent.

If some parameters are unused, commentout the variable name in the function definition:

// Always have named parameters ininterfaces.

class Shape {

 public:

 virtual void Rotate(double radians) = 0;

}

 

// Always have named parameters in thedeclaration.

class Circle : public Shape {

 public:

 virtual void Rotate(double radians);

}

 

// Comment out unused named parameters indefinitions.

void Circle::Rotate(double /*radians*/) {}

// Bad - if someone wants to implementlater, it's not clear what the

// variable means.

void Circle::Rotate(double) {}

Function Calls

link▽

On one line if it fits; otherwise, wraparguments at the parenthesis.

Function calls have the followingformat:

bool retval = DoSomething(argument1,argument2, argument3);

If the arguments do not all fit on oneline, they should be broken up onto multiple lines, with each subsequent linealigned with the first argument. Do not add spaces after the open paren orbefore the close paren:

bool retval =DoSomething(averyveryveryverylongargument1,

                          argument2,argument3);

If the function has many arguments,consider having one per line if this makes the code more readable:

bool retval = DoSomething(argument1,

                          argument2,

                          argument3,

                          argument4);

Arguments may optionally all be placedon subsequent lines, with one line per argument:

if (...) {

  ...

  ...

  if(...) {

   DoSomething(

       argument1,  // 4 space indent

       argument2,

       argument3,

       argument4);

  }

In particular, this should be done ifthe function signature is so long that it cannot fit within the maximum linelength.

Braced Initializer Lists

link▽

On one line if it fits, otherwise wrapat the open brace.

Put everything on one line wherepossible. If everything can't fit on one line, the open brace should be thelast character on its line, and the close brace should be the first characteron its line.

// Examples of braced init list on a singleline.

return {foo, bar};

functioncall({foo, bar});

pair<int, int> p{foo, bar};

 

// When you have to wrap.

MyType m = {

 superlongvariablename1,

 superlongvariablename2,

 {short, interior, list},

  {

   interiorwrappinglist,

   interiorwrappinglist2

  }

};

 

// Wrapping inside a function call.

function({

          wrapped, long,

          list, here

        });

 

// If the variable names are really long.

function(

    {

     wrapped,

     list

   });

Conditionals

link▽

Prefer no spaces inside parentheses. The else keyword belongs on a new line.

There are two acceptable formats for abasic conditional statement. One includes spaces between the parentheses andthe condition, and one does not.

The most common form is without spaces.Either is fine, but be consistent. If you are modifying a file, usethe format that is already present. If you are writing new code, use the formatthat the other files in that directory or project use. If in doubt and you haveno personal preference, do not add the spaces.

if (condition) {  // no spaces inside parentheses

 ...  // 2 space indent.

} else if (...) {  // The else goes on the same line as theclosing brace.

  ...

} else {

  ...

}

If you prefer you may add spaces insidethe parentheses:

if ( condition ) {  // spaces inside parentheses - rare

 ...  // 2 space indent.

} else { // The else goes on the same line as the closing brace.

  ...

}

Note that in all cases you must have aspace between the if and the open parenthesis. You mustalso have a space between the close parenthesis and the curly brace, if you'reusing one.

if(condition)     // Bad - space missing after IF.

if (condition){   // Bad - space missing before {.

if(condition){    // Doubly bad.

if (condition) {  // Good - proper space after IF and before {.

Short conditional statements may bewritten on one line if this enhances readability. You may use this only whenthe line is brief and the statement does not use the elseclause.

if (x == kFoo) return new Foo();

if (x == kBar) return new Bar();

This is not allowed when the ifstatement has an else:

// Not allowed - IF statement on one linewhen there is an ELSE clause

if (x) DoThis();

else DoThat();

In general, curly braces are notrequired for single-line statements, but they are allowed if you like them; conditionalor loop statements with complex conditions or statements may be more readablewith curly braces. Some projects require that an if must always always have an accompanying brace.

if (condition)

 DoSomething();  // 2 space indent.

 

if (condition) {

 DoSomething();  // 2 space indent.

}

However, if one part of an if-else statement uses curly braces, theother part must too:

// Not allowed - curly on IF but not ELSE

if (condition) {

 foo;

} else

 bar;

 

// Not allowed - curly on ELSE but not IF

if (condition)

 foo;

else {

 bar;

}

// Curly braces around both IF and ELSErequired because

// one of the clauses used braces.

if (condition) {

 foo;

} else {

 bar;

}

Loops and Switch Statements

link▽

Switch statements may use braces forblocks. Annotate non-trivial fall-through between cases. Empty loop bodiesshould use {} or continue.

case blocks in switch statements can have curly braces or not, dependingon your preference. If you do include curly braces they should be placed asshown below.

If not conditional on an enumeratedvalue, switch statements should always have a default case (in the case of an enumerated value, thecompiler will warn you if any values are not handled). If the default case shouldnever execute, simply assert:

switch (var) {

 case 0: {  // 2 space indent

   ...      // 4 space indent

   break;

  }

 case 1: {

   ...

   break;

  }

 default: {

   assert(false);

  }

}

Empty loop bodies should use {} or continue, but not asingle semicolon.

while (condition) {

  //Repeat test until it returns false.

}

for (int i = 0; i < kSomeNumber; ++i){}  // Good - empty body.

while (condition) continue;  // Good - continue indicates no logic.

while (condition);  // Bad - looks like part of do/while loop.

Pointer and Reference Expressions

link▽

No spaces around period or arrow.Pointer operators do not have trailing spaces.

The following are examples ofcorrectly-formatted pointer and reference expressions:

x = *p;

p = &x;

x = r.y;

x = r->y;

Note that:

• There are no spaces around the period or arrow when accessing a member.
• Pointer operators have no space after the * or &.

When declaring a pointer variable orargument, you may place the asterisk adjacent to either the type or to thevariable name:

// These are fine, space preceding.

char *c;

const string &str;

 

// These are fine, space following.

char* c;   // but remember to do "char* c, *d, *e, ...;"!

const string& str;

char * c; // Bad - spaces on both sides of *

const string & str;  // Bad - spaces on both sides of &

You should do this consistently within asingle file, so, when modifying an existing file, use the style in that file.

Boolean Expressions

link▽

When you have a boolean expression thatis longer than the standardline length, be consistent in how you break up the lines.

In this example, the logical AND operatoris always at the end of the lines:

if (this_one_thing > this_other_thing&&

   a_third_thing == a_fourth_thing &&

   yet_another && last_one) {

  ...

}

Note that when the code wraps in thisexample, both of the && logical AND operators are at theend of the line. This is more common in Google code, though wrapping alloperators at the beginning of the line is also allowed. Feel free to insertextra parentheses judiciously because they can be very helpful in increasingreadability when used appropriately. Also note that you should always use thepunctuation operators, such as && and ~, rather than the word operators, suchas and and compl.

Return Values

link▽

Do not needlessly surround the return expression with parentheses.

Use parentheses in return expr; only where you would use them in x = expr;.

return result;                  // No parentheses in thesimple case.

return (some_long_condition &&  // Parentheses ok to make a complex

       another_condition);     //     expression more readable.

return (value);                // You wouldn't write var =(value);

return(result);                // return is not a function!

Variable and Array Initialization

link▽

Your choice of =, (), or {}.

You may choose between =, (), and {}; the following are all correct:

int x = 3;

int x(3);

int x{3};

string name = "Some Name";

string name("Some Name");

string name{"Some Name"};

Be careful when using the {} on a type that takes an initializer_list in one of its constructors. The {} syntax prefers the initializer_list constructor whenever possible. To get the non- initializer_list constructor, use ().

vector<int> v(100, 1);  // A vector of 100 1s.

vector<int> v{100, 1};  // A vector of 100, 1.

Also, the brace form prevents narrowingof integral types. This can prevent some types of programming errors.

int pi(3.14);  // OK -- pi == 3.

int pi{3.14};  // Compile error: narrowing conversion.

Preprocessor Directives

link▽

The hash mark that starts a preprocessordirective should always be at the beginning of the line.

Even when preprocessor directives arewithin the body of indented code, the directives should start at the beginningof the line.

// Good - directives at beginning of line

  if(lopsided_score) {

#if DISASTER_PENDING      // Correct -- Starts at beginning of line

   DropEverything();

# if NOTIFY               // OK but not required -- Spacesafter #

   NotifyClient();

# endif

#endif

   BackToNormal();

  }

// Bad - indented directives

  if(lopsided_score) {

   #if DISASTER_PENDING  //Wrong!  The "#if" should be atbeginning of line

   DropEverything();

   #endif                //Wrong!  Do not indent "#endif"

   BackToNormal();

  }

Class Format

link▽

Sections in public, protected and private order,each indented one space.

The basic format for a class declaration(lacking the comments, see ClassComments for a discussion of what comments are needed) is:

class MyClass : public OtherClass {

 public:     // Note the 1 space indent!

 MyClass();  // Regular 2 spaceindent.

 explicit MyClass(int var);

 ~MyClass() {}

 

 void SomeFunction();

 void SomeFunctionThatDoesNothing() {

  }

 

 void set_some_var(int var) { some_var_ = var; }

  intsome_var() const { return some_var_; }

 

 private:

  boolSomeInternalFunction();

 

  intsome_var_;

  intsome_other_var_;

 DISALLOW_COPY_AND_ASSIGN(MyClass);

};

Things to note:

• Any base class name should be on the same line as the subclass name, subject to the 80-column limit.
• The public:, protected:, and private: keywords should be indented one space.
• Except for the first instance, these keywords should be preceded by a blank line. This rule is optional in small classes.
• Do not leave a blank line after these keywords.
• The public section should be first, followed by the protected and finally the private section.
• See Declaration Order for rules on ordering declarations within each of these sections.

Constructor Initializer Lists

link▽

Constructor initializer lists can be allon one line or with subsequent lines indented four spaces.

There are two acceptable formats forinitializer lists:

// When it all fits on one line:

MyClass::MyClass(int var) : some_var_(var),some_other_var_(var + 1) {}

or

// When it requires multiple lines, indent4 spaces, putting the colon on

// the first initializer line:

MyClass::MyClass(int var)

    :some_var_(var),             // 4 spaceindent

     some_other_var_(var + 1) {  //lined up

  ...

 DoSomething();

  ...

}

Namespace Formatting

link▽

The contents of namespaces are notindented.

Namespaces donot add an extra level of indentation. For example, use:

namespace {

 

void foo() {  // Correct. No extra indentation within namespace.

  ...

}

 

}  //namespace

Do not indent within a namespace:

namespace {

 

  //Wrong.  Indented when it should not be.

 void foo() {

   ...

  }

 

}  //namespace

When declaring nested namespaces, puteach namespace on its own line.

namespace foo {

namespace bar {

Horizontal Whitespace

link▽

Use of horizontal whitespace depends onlocation. Never put trailing whitespace at the end of a line.

General

void f(bool b) {  // Open braces should always have a spacebefore them.

  ...

int i = 0; // Semicolons usually have no space before them.

int x[] = { 0 };  // Spaces inside braces for braced-init-listare

int x[] = {0};    // optional.  If you use them, put them on both sides!

// Spaces around the colon in inheritanceand initializer lists.

class Foo : public Bar {

 public:

  //For inline function implementations, put spaces between the braces

  //and the implementation itself.

 Foo(int b) : Bar(), baz_(b) {}  //No spaces inside empty braces.

 void Reset() { baz_ = 0; }  //Spaces separating braces from implementation.

  ...

Adding trailing whitespace can causeextra work for others editing the same file, when they merge, as can removingexisting trailing whitespace. So: Don't introduce trailing whitespace. Removeit if you're already changing that line, or do it in a separate clean-upoperation (preferably when no-one else is working on the file).

Loops and Conditionals

if (b) {          // Space after the keyword inconditions and loops.

} else {          // Spaces around else.

}

while (test) {}   // There is usually no space insideparentheses.

switch (i) {

for (int i = 0; i < 5; ++i) {

switch ( i ) {    // Loops and conditions may have spacesinside

if ( test ) {     // parentheses, but this is rare.  Be consistent.

for ( int i = 0; i < 5; ++i ) {

for ( ; i < 5 ; ++i) {  // For loops always have a space after the

 ...                   //semicolon, and may have a space before the

                        // semicolon.

for (auto x : counts) {  // Range-based for loops always have a

 ...                    // spacebefore and after the colon.

}

switch (i) {

 case 1:         // No space beforecolon in a switch case.

   ...

 case 2: break;  // Use a spaceafter a colon if there's code after it.

Operators

x = 0;              // Assignment operators alwayshave spaces around

                    // them.

x = -5;             // No spaces separating unaryoperators and their

++x;                // arguments.

if (x && !y)

  ...

v = w * x + y / z;  // Binary operators usually have spacesaround them,

v = w*x + y/z;      // but it's okay to remove spaces aroundfactors.

v = w * (x + z);    // Parentheses should have no spaces insidethem.

Templates and Casts

vector<string> x;           // No spaces inside the angle

y = static_cast<char*>(x);  // brackets (< and >), before

                            // <, or between>( in a cast.

vector<char *> x;           // Spaces between type and pointerare

                            // okay, but be consistent.

set<list<string>> x;        // Permitted in C++11 code.

set<list<string> > x;       // C++03 required a space in > >.

set< list<string> > x;      // You may optionally use

                            // symmetric spacingin < <.

Vertical Whitespace

link▽

Minimize use of vertical whitespace.

This is more a principle than a rule:don't use blank lines when you don't have to. In particular, don't put morethan one or two blank lines between functions, resist starting functions with ablank line, don't end functions with a blank line, and be discriminating withyour use of blank lines inside functions.

The basic principle is: The more codethat fits on one screen, the easier it is to follow and understand the controlflow of the program. Of course, readability can suffer from code being toodense as well as too spread out, so use your judgement. But in general,minimize use of vertical whitespace.

Some rules of thumb to help when blanklines may be useful:

• Blank lines at the beginning or end of a function very rarely help readability.
• Blank lines inside a chain of if-else blocks may well help readability.

Exceptions to the Rules

The coding conventions described aboveare mandatory. However, like all good rules, these sometimes have exceptions,which we discuss here.

Existing Non-conformant Code

link▽

You may diverge from the rules whendealing with code that does not conform to this style guide.

If you find yourself modifying code thatwas written to specifications other than those presented by this guide, you mayhave to diverge from these rules in order to stay consistent with the localconventions in that code. If you are in doubt about how to do this, ask theoriginal author or the person currently responsible for the code. Remember that consistency includeslocal consistency, too.

Windows Code

link▽

Windows programmers have developed theirown set of coding conventions, mainly derived from the conventions in Windowsheaders and other Microsoft code. We want to make it easy for anyone tounderstand your code, so we have a single set of guidelines for everyonewriting C++ on any platform.

It is worth reiterating a few of theguidelines that you might forget if you are used to the prevalent Windowsstyle:

• Do not use Hungarian notation (for example, naming an integer iNum). Use the Google naming conventions, including the .cc extension for source files.
• Windows defines many of its own synonyms for primitive types, such as DWORD, HANDLE, etc. It is perfectly acceptable, and encouraged, that you use these types when calling Windows API functions. Even so, keep as close as you can to the underlying C++ types. For example, use const TCHAR * instead of LPCTSTR.
• When compiling with Microsoft Visual C++, set the compiler to warning level 3 or higher, and treat all warnings as errors.
• Do not use #pragma once; instead use the standard Google include guards. The path in the include guards should be relative to the top of your project tree.
• In fact, do not use any nonstandard extensions, like #pragma and __declspec, unless you absolutely must. Using __declspec(dllimport) and __declspec(dllexport) is allowed; however, you must use them through macros such as DLLIMPORT and DLLEXPORT, so that someone can easily disable the extensions if they share the code.

However, there are just a few rules thatwe occasionally need to break on Windows:

• Normally we forbid the use of multiple implementation inheritance; however, it is required when using COM and some ATL/WTL classes. You may use multiple implementation inheritance to implement COM or ATL/WTL classes and interfaces.
• Although you should not use exceptions in your own code, they are used extensively in the ATL and some STLs, including the one that comes with Visual C++. When using the ATL, you should define _ATL_NO_EXCEPTIONS to disable exceptions. You should investigate whether you can also disable exceptions in your STL, but if not, it is OK to turn on exceptions in the compiler. (Note that this is only to get the STL to compile. You should still not write exception handling code yourself.)
• The usual way of working with precompiled headers is to include a header file at the top of each source file, typically with a name like StdAfx.h or precompile.h. To make your code easier to share with other projects, avoid including this file explicitly (except in precompile.cc), and use the /FI compiler option to include the file automatically.
• Resource headers, which are usually named resource.h and contain only macros, do not need to conform to these style guidelines.

Parting Words

Use common sense and BECONSISTENT.

If you are editing code, take a fewminutes to look at the code around you and determine its style. If they usespaces around their if clauses, you should, too. If theircomments have little boxes of stars around them, make your comments have littleboxes of stars around them too.

The point of having style guidelines isto have a common vocabulary of coding so people can concentrate on what you aresaying, rather than on how you are saying it. We present global style ruleshere so people know the vocabulary. But local style is also important. If codeyou add to a file looks drastically different from the existing code around it,the discontinuity throws readers out of their rhythm when they go to read it.Try to avoid this.

OK, enough writing about writing code;the code itself is much more interesting. Have fun!

---

Revision 3.260

Benjy Weinberger
Craig Silverstein
Gregory Eitzmann
Mark Mentovai
Tashana Landray