Difference between revisions of "CodingStandards"

From The Battle for Wesnoth Wiki
(C++ version)
(48 intermediate revisions by 20 users not shown)
Line 1: Line 1:
Wesnoth uses modern/advanced C++ that is portable to Visual C++ 6 and GNU G++ 3.0+
+
Wesnoth uses C++ that is portable to C++ compilers targeting various commonly used platforms.
  
== End Non-Public Members of Classes with an Underscore ==
+
== C++ version ==
  
All non-public data members of classes should have their names terminated with an underscore, to show that they are a
+
Wesnoth uses C++ conforming to C++11 in stable branch and C++14 in master branch.
class member. This makes for more readable code, once one is familiar with the convention.
 
  
== Use References when a value may not be NULL ==
+
== Formatting ==
  
If a value passed to a function can never be NULL, use a reference instead of a pointer. I.e.
+
When working on C++ for Wesnoth, indent your code with a tab character. After fully indenting, if you still need to line up the text with a specific character on the line above, you may further align it using space characters.
  
  void myfunction(Object& obj);
+
You may use long lines.
 +
 
 +
Don't use a space after control flow keywords (if/for/...).
 +
 
 +
== Evil things to avoid ==
 +
 
 +
=== Avoid implicit conversions ===
 +
 
 +
Make all constructors which only take one argument that is of a different type to the class <tt>explicit</tt>.
 +
 
 +
Do not use <tt>operator T()</tt> (where <tt>T</tt> is a type) to allow an implicit conversion to a different type. For example:
 +
 
 +
t_string(const std::string&);
 +
 
 +
This can cause many situations where a temporary t_string is implicitly created and then gets destroyed unexpectedly (reference [https://gna.org/bugs/index.php?20360 bug #20360]).
 +
 
 +
=== Do not declare class data members as non-private ===
 +
 
 +
It's okay to have a ''struct'' with only public members, if that's what you want.
 +
 
 +
However, once something is a ''class'' with private data members, do not add public (or even protected) data members to the class. Doing this breaks encapsulation and can cause all kinds of confusing and evil things to happen.
 +
 
 +
=== Destructors must not throw exceptions ===
 +
 
 +
Do not allow exceptions to propogate from a destructor. Doing that is always bad in C++. Any code which does it should be treated as a bug and fixed. Doing this is a very easy way to cause memory leaks and crashes.
 +
 
 +
It's okay to have exceptions thrown inside a destructor, just as long as you catch() them inside the destructor too and don't allow them to propagate out.
 +
 
 +
In C++11, any destructor which throws an exception causes the program to be immediately terminated (except in special cases).
 +
 
 +
You can read more about the issue here: http://c2.com/cgi/wiki?BewareOfExceptionsInTheDestructor
 +
 
 +
== Naming ==
 +
 
 +
=== End non-public class data members with an underscore ===
 +
 
 +
All non-public data members of classes should have their names terminated with an underscore, to show that they are a class member. This makes for more readable code, once one is familiar with the convention.
 +
 
 +
== Idioms ==
 +
 
 +
=== Use references when a value may not be NULL ===
 +
 
 +
If a value passed to a function can never be <tt>NULL</tt>, use a reference instead of a pointer. For example:
 +
 
 +
void my_function(T& obj);
  
 
rather than
 
rather than
  
  void myfunction(Object* obj);
+
void my_function(T* obj);
 +
 
 +
This more clearly shows prospective users of the function that <tt>obj</tt> may never be <tt>NULL</tt>, without them having to consult documentation or the implementation of the function.
 +
 
 +
=== Use the const keyword ===
  
This more clearly shows the user of the function that obj may never be NULL, without them having to consult
+
The <tt>const</tt> keyword in C++ allows interfaces to more clearly specify how they treat objects.
documentation or the implementation of the function.
+
Always use <tt>const</tt> when you are not going to modify an object. For example:
  
== Use Const ==
+
void my_function(const T& obj);
  
The 'const' feature of C++ allows interfaces to more clearly specify how they treat objects. Always use const when you
+
This shows to the caller that <tt>obj</tt> will not be modified. If <tt>my_function()</tt> may modify <tt>obj</tt>, then use the following instead:
are not going to modify an object.
 
  
I.e.
+
void my_function(T& obj);
  
  void myfunction(const Object& obj);
+
Likewise, if a variable is not changed after initialization, make it <tt>const</tt>, and mark member functions as <tt>const</tt> if they do not modify their object.
  
demonstrates to the caller of myfunction() that obj will not be modified. If myfunction may modify obj, then use
+
=== Know the behavior of const references when types differ ===
  
  void myfunction(Object& obj);
+
If you assign something to a <tt>const</tt> reference of a different type, if necessary (if the type is different but there is a conversion) the compiler will create a temporary and guarantee it lasts for the lifetime of the reference. So
  
likewise, if a variable is not changed after initialization, make it const.
+
char c = 0;
 +
const int& i = c;
 +
c = 5;
  
== Write Exception-Safe Code ==
+
will result in c == 5 and i == 0, which may not be what you expect.
 +
 
 +
=== Write exception-safe code ===
  
 
Wesnoth code should be exception-safe, even if you do not use exceptions directly. That is, you should be able to assume that an exception is thrown almost anywhere from within the code, with well-defined results (i.e. no resource leaks).
 
Wesnoth code should be exception-safe, even if you do not use exceptions directly. That is, you should be able to assume that an exception is thrown almost anywhere from within the code, with well-defined results (i.e. no resource leaks).
  
Code that uses a pattern like,
+
Code that uses a pattern like the following is bad:
  
  {
+
{
  SDL_Surface* image = IMG_Load("image.bmp");
+
    SDL_Surface* image = IMG_Load("image.bmp");
  ...some code, which uses 'image'...
+
    ...some code, which uses 'image'...
  SDL_FreeSurface(image);
+
    SDL_FreeSurface(image);
  }
+
}
  
is bad, because the code may throw an exception, and 'image' will never be freed. Instead, use wrapper objects which free the object in their destructor.
+
The code may throw an exception, and <tt>image</tt> will never be freed. Instead, use wrapper objects which free the object in their destructor.
  
For SDL_Surface objects, use the <tt>surface</tt> class. So you could rewrite the above code,
+
For <tt>SDL_Surface</tt> objects, the <tt>surface</tt> type is used throughout the Wesnoth source code to achieve this purpose. So you could rewrite the above code as follows:
  
  {
+
{
  surface image(IMG_Load("image.bmp"));
+
    surface image(IMG_Load("image.bmp"));
  ...some code, which uses 'image'...
+
    ...some code, which uses 'image'...
  } ''the image is automatically freed here when 'image' is destroyed
+
} ''the image is automatically freed here when 'image' is destroyed
  
Instead of allocating memory directly using new[] or malloc(), use language-provided containers, such as vector.
+
Instead of allocating memory directly using <tt>new[]</tt> or <tt>malloc()</tt>, use language-provided containers, such as vector.
  
== Respect for loop scoping of different platforms ==
+
Similarly, avoid writing <tt>delete</tt> explicitly in your code. Oftentimes using a smart pointer or similar instead will improve the readability of your code, by making it obvious that you are managing memory correctly even if an exception is thrown. If you were deleting a pointer which is a member variable of an object, using a smart pointer instead may simplify your code by eliminating the need to write an explicit destructor in some cases.
  
In the code,
+
For more information, you can read about the (important) RAII idiom: http://c2.com/cgi/wiki?ResourceAcquisitionIsInitialization
  
  for(int i = 0; i != 100; ++i) {...}
+
=== Do not use sprintf ===
  
the variable 'i' is scoped within the for loop according to ISO/ANSI C++ and GNU G++. However it is scoped within the
+
The <tt>sprintf()</tt> function does not check whether or not it is writing past the end of the space allocated.  This is a security problem if someone other than the person running the game can cause <tt>sprintf()</tt> to write very long strings. In Wesnoth, this untrusted data could come potentially from other players in a multiplayer game, or from downloaded add-ons. Instead you should use <tt>snprintf()</tt> with the second argument being the <tt>sizeof</tt> of the buffer that will hold the result.
surrounding scope according to Visual C++ 6.
 
  
This means that the code,
+
== Standard C++ to avoid ==
  
  for(int i = 0; i != 100; ++i) {}
+
=== Do not use 0 or NULL when you mean nullptr ===
  for(int i = 0; i != 100; ++i) {}
 
  
is illegal on VC++6, because i is defined twice, although it is legal according to the standard, and GNU G++.
+
Several Wesnoth developers, including Dave, find the number 0 to be very ambiguous when used in a non-numeric context. In keeping with the precedent that has already been established in the Wesnoth source code, you should avoid using literal zero for initializing and/or comparing null pointers.
  
On VC++6, the legal way to write it would be,
+
=== Do not use standard io functions ===
 +
This implies the std::fstream class and fopen. These functions do not support utf8 in windows which is our standard encoding for filepaths. Use our custom filesystem functions (filesystem.hpp) instead.
  
  for(int i = 0; i != 100; ++i) {}
+
== C legacy to be avoided ==
  for(i = 0; i != 100; ++i) {}
 
  
But this is illegal according to the standard, because 'i' is not defined in the second loop. The correct way to write
+
=== Use std::array instead of C-style Arrays ===
this code to conform to the standard and work on all platforms is to simply abandon declaring variables in the
 
initialization statement of a for loop when the variable must be reused in the same scope,
 
  
  int i;
+
C-style arrays are very efficient, but their interface is ugly. Use <tt>std::array</tt> instead.
  for(i = 0; i != 100; ++i) {}
 
  for(i = 0; i != 100; ++i) {}
 
  
== Use the function templates minimum and maximum ==
+
=== Do not use C-style or function-style casts ===
  
Standard C++ offers the function templates min and max to find the minimum and maximum of two values on which operator
+
The following code,
<
 
is defined. Unfortunately, many hoops must be leapt through to get this working on VC++. So, we do not use standard
 
min
 
and max. Instead, we use minimum and maximum, defined in utils.hpp.
 
  
Usage is fairly natural:
+
if(i->second.side() == (size_t)player_number_) {
  
  int i = minimum(x,y);
+
is considered bad practice in C++ since a C-style cast is overpowered -- if types change around it could end up casting away constness, or performing an implementation-defined data reinterpretation (basically a C-style cast is a compiler-generated combination of <tt>static_cast</tt>, <tt>reinterpret_cast</tt>, and <tt>const_cast</tt>).
  
Note that in the above example, if x is an unsigned integer, and y is a signed integer, VC++ will have problems. You
+
This syntax,
must explicitly specify the version of minimum being called in such cases:
 
  
  int i = minimum<int>(x,y);
+
if(i->second.side() == size_t(player_number_)) {
  
== Do not use wstring ==
+
is also a cast, and [http://stackoverflow.com/a/32224 semantically identical] to the C-style cast. It should also be avoided.
  
The standard C++ wstring class, defined as a basic_string< wchar_t >, does not exist in some platforms supported
+
Good programming style is to use the least powerful tool available that does what you want. For example:
by
 
Wesnoth. Use wide_string, defined in language.hpp, instead. wide_string is actually defined as a vector< wchar_t >
 
  
== Use util::array instead of C-style Arrays ==
+
if(i->second.side() == static_cast<size_t>(player_number_)) {
  
C-style arrays are very efficient, but their interface is ugly. Use util::array defined in array.hpp. It is a wrapper
+
Alternatively, a constructor call may be used for non-built-in types.
for an array which has a C++ container-style interface. If you need to, extend it to make it fit your needs.
 
  
== Do not use sprintf ==
+
''Note: there may be some obscure cases where a C-style cast is desirable, such as converting a pointer to an integer type of unspecified size.''
  
Sprintf does not check whether or not it is writing passed the end of the space allocated. This is a security problem if someone other than the person running the game can cause sprintf to write very long strings. In Wesnoth this untrusted data could come potentially from other players in a multiplayer game or from downloaded campaigns. Instead you should use snprintf with the second argument being sizeof of the buffer that will hold the result.
+
=== Do not use #define for constants ===
  
== Do not use C-style casts ==
+
<tt><nowiki>#</nowiki>define foo X</tt> is not a typesafe approach to define constants. Instead, you can something like the following (in an anonymous namespace) to achieve the same goal in a typesafe fashion.
  
The following code,
+
namespace {
 +
    const T foo = X;
 +
}
 +
 
 +
== Documentation ==
  
  if(i->second.side() == (size_t)player_number_) {
+
=== Document config preconditions and postconditions ===
  
is considered bad practice in C++ since a C-style cast is overpowered -- if types change around it could end up casting away constness, or performing an implementation-defined data reinterpretation (basically a C-style cast is a compiler generated combination of static_cast, reinterpret_cast, and const_cast).
+
In the Wesnoth code you will commonly encounter a data container type known as <tt>config</tt>, which contains hierarchical string data (such as WML contents or game settings). The tagged ''children'' of the <tt>config</tt> object and their string ''attributes'' are arranged in an ordered and mapped format, internally implemented using the C++ STL.
  
Good programming style is to use the least powerful tool available that does what you want. For example,
+
Because <tt>config</tt> data is utilized in so many ways and places, it can be difficult to track across the scope of the entire program. Thus, you should document all public functions that take/return <tt>config</tt> objects, specifying content expectations and updating any related entries in the [[ReferenceWML]] wiki pages. In particular, if your function requires a <tt>config</tt> parameter, specify where/how the <tt>config</tt> object should be created. This will be a great help to any future coders who need to call or modify your function.
  
  if(i->second.side() == static_cast<size_t>(player_number_)) {
+
=== Doxygen ===
  
Alternatively, a constructor call may be used for non-builtin types.
+
See [[Doxygen]] for tips on how to comment the code, so that Doxygen can nicely document it.
  
== Document "config" preconditions and postconditions ==
+
== See also ==
  
In the Wesnoth code you will commonly encounter a data container known as the "config," which contains heirarchical string data (such as WML contents or game settings). The tagged "children" of the config and their string "attributes" are arranged in an ordered and mapped format internally using STL.
+
* [[HackingWesnoth]]
  
Because config data is utilized in so many ways and places,  it can be difficult to track across the scope of the entire program. You should document all public functions that take/return a config, specifying config content expectations. In particular, if your function requires a config parameter, specify where/how the config should be created. This will be a great help to any future coders who need to call or modify your function.
+
[[Category:Development]]

Revision as of 21:17, 24 January 2019

Wesnoth uses C++ that is portable to C++ compilers targeting various commonly used platforms.

C++ version

Wesnoth uses C++ conforming to C++11 in stable branch and C++14 in master branch.

Formatting

When working on C++ for Wesnoth, indent your code with a tab character. After fully indenting, if you still need to line up the text with a specific character on the line above, you may further align it using space characters.

You may use long lines.

Don't use a space after control flow keywords (if/for/...).

Evil things to avoid

Avoid implicit conversions

Make all constructors which only take one argument that is of a different type to the class explicit.

Do not use operator T() (where T is a type) to allow an implicit conversion to a different type. For example:

t_string(const std::string&);

This can cause many situations where a temporary t_string is implicitly created and then gets destroyed unexpectedly (reference bug #20360).

Do not declare class data members as non-private

It's okay to have a struct with only public members, if that's what you want.

However, once something is a class with private data members, do not add public (or even protected) data members to the class. Doing this breaks encapsulation and can cause all kinds of confusing and evil things to happen.

Destructors must not throw exceptions

Do not allow exceptions to propogate from a destructor. Doing that is always bad in C++. Any code which does it should be treated as a bug and fixed. Doing this is a very easy way to cause memory leaks and crashes.

It's okay to have exceptions thrown inside a destructor, just as long as you catch() them inside the destructor too and don't allow them to propagate out.

In C++11, any destructor which throws an exception causes the program to be immediately terminated (except in special cases).

You can read more about the issue here: http://c2.com/cgi/wiki?BewareOfExceptionsInTheDestructor

Naming

End non-public class data members with an underscore

All non-public data members of classes should have their names terminated with an underscore, to show that they are a class member. This makes for more readable code, once one is familiar with the convention.

Idioms

Use references when a value may not be NULL

If a value passed to a function can never be NULL, use a reference instead of a pointer. For example:

void my_function(T& obj);

rather than

void my_function(T* obj);

This more clearly shows prospective users of the function that obj may never be NULL, without them having to consult documentation or the implementation of the function.

Use the const keyword

The const keyword in C++ allows interfaces to more clearly specify how they treat objects. Always use const when you are not going to modify an object. For example:

void my_function(const T& obj);

This shows to the caller that obj will not be modified. If my_function() may modify obj, then use the following instead:

void my_function(T& obj);

Likewise, if a variable is not changed after initialization, make it const, and mark member functions as const if they do not modify their object.

Know the behavior of const references when types differ

If you assign something to a const reference of a different type, if necessary (if the type is different but there is a conversion) the compiler will create a temporary and guarantee it lasts for the lifetime of the reference. So

char c = 0;
const int& i = c;
c = 5;

will result in c == 5 and i == 0, which may not be what you expect.

Write exception-safe code

Wesnoth code should be exception-safe, even if you do not use exceptions directly. That is, you should be able to assume that an exception is thrown almost anywhere from within the code, with well-defined results (i.e. no resource leaks).

Code that uses a pattern like the following is bad:

{
    SDL_Surface* image = IMG_Load("image.bmp");
    ...some code, which uses 'image'...
    SDL_FreeSurface(image);
}

The code may throw an exception, and image will never be freed. Instead, use wrapper objects which free the object in their destructor.

For SDL_Surface objects, the surface type is used throughout the Wesnoth source code to achieve this purpose. So you could rewrite the above code as follows:

{
    surface image(IMG_Load("image.bmp"));
    ...some code, which uses 'image'...
} the image is automatically freed here when 'image' is destroyed

Instead of allocating memory directly using new[] or malloc(), use language-provided containers, such as vector.

Similarly, avoid writing delete explicitly in your code. Oftentimes using a smart pointer or similar instead will improve the readability of your code, by making it obvious that you are managing memory correctly even if an exception is thrown. If you were deleting a pointer which is a member variable of an object, using a smart pointer instead may simplify your code by eliminating the need to write an explicit destructor in some cases.

For more information, you can read about the (important) RAII idiom: http://c2.com/cgi/wiki?ResourceAcquisitionIsInitialization

Do not use sprintf

The sprintf() function does not check whether or not it is writing past the end of the space allocated. This is a security problem if someone other than the person running the game can cause sprintf() to write very long strings. In Wesnoth, this untrusted data could come potentially from other players in a multiplayer game, or from downloaded add-ons. Instead you should use snprintf() with the second argument being the sizeof of the buffer that will hold the result.

Standard C++ to avoid

Do not use 0 or NULL when you mean nullptr

Several Wesnoth developers, including Dave, find the number 0 to be very ambiguous when used in a non-numeric context. In keeping with the precedent that has already been established in the Wesnoth source code, you should avoid using literal zero for initializing and/or comparing null pointers.

Do not use standard io functions

This implies the std::fstream class and fopen. These functions do not support utf8 in windows which is our standard encoding for filepaths. Use our custom filesystem functions (filesystem.hpp) instead.

C legacy to be avoided

Use std::array instead of C-style Arrays

C-style arrays are very efficient, but their interface is ugly. Use std::array instead.

Do not use C-style or function-style casts

The following code,

if(i->second.side() == (size_t)player_number_) {

is considered bad practice in C++ since a C-style cast is overpowered -- if types change around it could end up casting away constness, or performing an implementation-defined data reinterpretation (basically a C-style cast is a compiler-generated combination of static_cast, reinterpret_cast, and const_cast).

This syntax,

if(i->second.side() == size_t(player_number_)) {

is also a cast, and semantically identical to the C-style cast. It should also be avoided.

Good programming style is to use the least powerful tool available that does what you want. For example:

if(i->second.side() == static_cast<size_t>(player_number_)) {

Alternatively, a constructor call may be used for non-built-in types.

Note: there may be some obscure cases where a C-style cast is desirable, such as converting a pointer to an integer type of unspecified size.

Do not use #define for constants

#define foo X is not a typesafe approach to define constants. Instead, you can something like the following (in an anonymous namespace) to achieve the same goal in a typesafe fashion.

namespace {
    const T foo = X;
}

Documentation

Document config preconditions and postconditions

In the Wesnoth code you will commonly encounter a data container type known as config, which contains hierarchical string data (such as WML contents or game settings). The tagged children of the config object and their string attributes are arranged in an ordered and mapped format, internally implemented using the C++ STL.

Because config data is utilized in so many ways and places, it can be difficult to track across the scope of the entire program. Thus, you should document all public functions that take/return config objects, specifying content expectations and updating any related entries in the ReferenceWML wiki pages. In particular, if your function requires a config parameter, specify where/how the config object should be created. This will be a great help to any future coders who need to call or modify your function.

Doxygen

See Doxygen for tips on how to comment the code, so that Doxygen can nicely document it.

See also