Difference between revisions of "Coding guidelines"

From VCMI Project Wiki
Jump to: navigation, search
(C++ Standard)
(Coding Guidelines)
Line 19: Line 19:
 
* forward enum declarations,
 
* forward enum declarations,
 
* strongly-typed enums.
 
* strongly-typed enums.
 
 
=== New features in MSVC 2013 ===
 
=== New features in MSVC 2013 ===
 
In future VCMI may require installation of MSVC 2013. This would allow to use:
 
In future VCMI may require installation of MSVC 2013. This would allow to use:
Line 29: Line 28:
 
=== New features in GCC 4.7 ===
 
=== New features in GCC 4.7 ===
 
VCMI will require gcc-4.7 once it will become available in majority of Linux distributions. This would allow to use:
 
VCMI will require gcc-4.7 once it will become available in majority of Linux distributions. This would allow to use:
 +
* override and final.
 +
=== New features in GCC 4.7 and MSCV ===
 
* non-static data member initializers,
 
* non-static data member initializers,
* override and final,
+
* extended friend Declarations,
 
* template aliases.
 
* template aliases.
=== New features in GCC 4.7 and MSCV ===
 
* Extended friend Declarations.
 
 
 
== Style Guidelines ==
 
== Style Guidelines ==
  

Revision as of 11:50, 4 March 2014

Coding Guidelines

C++ Standard

VCMI implementation bases on C++03 standard with several extensions available in compilers we target (GCC 4.6+, MSVC 2012+, Clang 3.1+)

Further information about these extensions and compiler support for them is available at [1].

Currently allowed C++11 features

  • auto keyword,
  • decltype,
  • lambda expressions and closures,
  • local and unnamed types as template arguments,
  • new function declaration syntax for deduced return types (returned type after argument list),
  • right angle brackets,
  • r-value references and std::move,
  • static assert,
  • nullptr keyword,
  • range-based for loops,
  • forward enum declarations,
  • strongly-typed enums.

New features in MSVC 2013

In future VCMI may require installation of MSVC 2013. This would allow to use:

  • explicit conversion operators,
  • initializer lists,
  • raw string literals,
  • variadic templates,
  • defaulted and deleted functions (except for defaulted move functions).

New features in GCC 4.7

VCMI will require gcc-4.7 once it will become available in majority of Linux distributions. This would allow to use:

  • override and final.

New features in GCC 4.7 and MSCV

  • non-static data member initializers,
  • extended friend Declarations,
  • template aliases.

Style Guidelines

In order to keep the code consistent, please use the following conventions. From here on `good' and `bad' are used to attribute things that would make the coding style match, or not match. It is not a judgment call on your coding abilities, but more of a style and look call. Please try to follow these guidelines to ensure prettiness.


Indentation

Use 4 space tabs for writing your code (hopefully we can keep this consistent). If you are modifying someone else's code, try to keep the coding style similar. Switch statements have the case at the same indentation as the switch.

Switch statement

switch(alignment)
{
case EAlignment::EVIL:
    do_that();
    break;
case EAlignment::GOOD:
   do_that();
   break;
}

Where to put spaces

Use a space before and after the address or pointer character in a pointer declaration.

Good:

CIntObject * images[100];

Bad:

CIntObject* images[100]; or
CIntObject *images[100];


Use whitespace for clarity

Use white space in expressions liberally, except in the presence of parenthesis.

Good:

if(a + 5 > method (blah ('a') + 4))
    foo += 24;

Bad:

if(a+5>method(blah('a')+4))
foo+=24;

Between if, for, while,.. and the opening brace there shouldn't be a whitespace. The keywords are highlighted, so they don't need further separation.

Where to put braces

Inside a code block put the opening brace on the next line after the current statement:

Good:

if(a) 
{
    code ();
    code ();
}

Bad:

if(a) {
    code ();
    code ();
}

Avoid using unnecessary open/close braces, vertical space is usually limited:

Good:

if(a)
    code ();

Better:

if(a) code ();

Bad:

if(a) {
    code ();
}

When defining a method, use a new line for the brace, like this:

Good:

void method()
{
}

Bad:

void Method() {
}

When allocating objects, don't use parentheses for creating stack-based objects by zero param c-tors to avoid c++ most vexing parse and use parentheses for creating heap-based objects.

Good:

std::vector<int> v; 
CGBoat btn = new CGBoat();

Bad:

std::vector<int> v(); // shouldn't compile anyway 
CGBoat btn = new CGBoat;

File headers

For any new files, please use a descriptive introduction, like this:

/*
 * Name_of_File.h, part of VCMI engine
 *
 * Authors: listed in file AUTHORS in main folder
 *
 * License: GNU General Public License v2.0 or later
 * Full text of license available in license.txt file, in main folder
 *
 */

The above notice have to be included only in header files (.h), except there is a cpp file with no corresponding header file.

Where and how to comment

You should write a comment before the class definition which describes shortly the class. 1-2 sentences are enough. Methods and class data members should be commented if they aren't self-describing only. Getters/Setters, simple methods where the purpose is clear or similar methods shouldn't be commented, because vertical space is usually limited. The style of documentation comments should be the three slashes-style: ///.

/// Returns true if a debug/trace log message will be logged, false if not.
/// Useful if performance is important and concatenating the log message is a expensive task.
bool isDebugEnabled() const;
bool isTraceEnabled() const;

The above example doesn't follow a strict scheme on how to comment a method. It describes two methods in one go. Comments should be kept short.

A good essay about writing comments: [2]

Casing

Local variables and methods start with a lowercase letter and use the camel casing. Classes/Structs start with an uppercase letter and use the camel casing as well. Macros and constants are written uppercase.


Line length

The line length for c++ source code is 120 columns. If your function declaration arguments go beyond this point, please align your arguments to match the opening brace. For best results use the same number of tabs used on the first line followed by enough spaces to align the arguments.


Warnings

Avoid use of #pragma to disable warnings. Compile at warning level 3. Avoid commiting code with new warnings.

Type naming

Classes are prefixed with an upper C, interfaces with an upper I, enumerations with an upper E, structs without an prefix and typedefs with an upper T

File/directory naming

Compilation units(.cpp,.h files) start with a uppercase letter and are named like the name of a class which resides in that file if possible. Header only files start with a uppercase letter. JSON files start with a lowercase letter and use the camel casing.

Directories start with a lowercase letter and use the camel casing where necessary.


Logging

If you want to trace the control flow of VCMI, then you should use the macro LOG_TRACE or LOG_TRACE_PARAMS. The first one prints a message when the function is entered or leaved. The name of the function will also be logged. In addition to this the second macro, let's you specify parameters which you want to print. You should print traces with parameters like this:

LOG_TRACE_PARAMS(logGlobal, "hero '%s', spellId '%d', pos '%s'.", hero, spellId, pos);

When using the macro every "simple" parameter should be logged. The parameter can be a number, a string or a type with a ostream operator<<. You should not log contents of a whole text file, a byte array or sth. like this. If there is a simple type with a few members you want to log, you should write an ostream operator<<. The produced message can look like this:

{BattleAction: side '0', stackNumber '1', actionType 'Walk and attack', destinationTile '{BattleHex: x '7', y '1', hex '24'}', additionalInfo '7', selectedStack '-1'}

The name of the type should be logged first, e.g. {TYPE_NAME: members...}. The members of the object will be logged like logging trace parameters. Collection types (vector, list, ...) should be logged this way: [{BattleHex: ...}, {...}] There is no format which has to be followed strictly, so if there is a reason to format members/objects in a different way, then this is ok.

Best practices

Avoid code duplication

Avoid code duplication or don't repeat yourself(DRY) is the most important aspect in programming. Code duplication of any kind can lead to inconsistency and is much harder to maintain. If one part of the system gets changed you have to change the code in several places. This process is error-prone and leads often to problems. Here you can read more about the DRY principle: http://en.wikipedia.org/wiki/Don%27t_repeat_yourself

Loop handling

Use BOOST_FOREACH for iterating through every item of a container. It should be used in any case except if Visual Studio debug performance is important, then you may use a simple for loop to avoid use of STL Iterator checks.

The loop counter should be of type int, unless you are sure you won't need negative indices -- then use size_t.

Include guards

Use #pragma once instead of the traditional #ifndef/#define/#endif include guards.


Pre compiled header file

The header StdInc.h should be included in every compilation unit. It has to be included before any C macro and before any c++ statements. Pre compiled header should not be changed, except any important thing is missing. The StdInc includes most Boost libraries and nearly all standard STL and C libraries, so you don’t have to include them by yourself.


Enumeration handling

Do not declare enumerations in global namespace. It is better to wrap them in class or namespace to avoid polluting global namespace:

namespace EAlignment
{
	enum EAlignment { GOOD, EVIL, NEUTRAL };
}

Avoid senseless comments

If the comment duplicates the name of commented member, it's better if it wouldn't exist at all. It just increases maintenance cost. Bad:

size_t getHeroesCount(); //gets count of heroes (surprise?)


Class handling

There is no definitive rule which has to be followed strictly. You can freely decide if you want to pack your own classes, where you are programming on, all in one file or each in one file. It's more important that you feel comfortable with the code, than consistency overall the project. VCMI has several container class files, so if you got one additional class to them than just add it to them instead of adding new files.

Functions and interfaces

Don't return const objects or primitive types from functions -- it's pointless. Also, don't return pointers to non-const game data objects from callbacks to player interfaces.

Bad:

const std::vector<CGObjectInstance*> guardingCreatures (int3 pos) const;

Good:

std::vector<const CGObjectInstance*> guardingCreatures (int3 pos) const;

Sources

Mono project coding guidelines