Open main menu

Code Formatting Conventions

Revision as of 17:20, 27 December 2011 by Clone2727 (talk | contribs) (→‎Switch / Case constructs: continue with camel case here too)

Use common sense

These are conventions which we try to follow when writing code for ScummVM. Sticking to a common set of formatting / indention rules makes it easier to read through our source base (which now exceed half a million lines of code by far, making this quite important). If you want to submit patches, please try to adhere to these rules.

We don't always follow these rules slavishly, in certain cases it is OK (and in fact might be favorable) to stray from them. Applying common sense, as usual, is a good idea.

In the following examples tabs are replaced by spaces for visual consistency with the Code Formatting Conventions. But in actual code, use tabs for indentations (our tabs are assumed to be 4 spaces wide).

Hugging braces

Braces in your code should look like the following example:

<syntax type="C++"> for (int i = 0; i < t; i++) {

   [...]

}

if (j < k) {

   [...]

} else if (j > k) {

   [...]

} else {

   [...]

}

class Dummy {

   [...]

}; </syntax>

Did you see the {}'s on that?

Tab indents, with tabstop at four spaces

Says it all, really.

Whitespaces

Conventional operators surrounded by a space character

<syntax type="C++"> a = (b + c) * d; </syntax>

C++ reserved words separated from opening parentheses by a white space

<syntax type="C++"> while (true) { </syntax>

Commas followed by a white space

<syntax type="C++"> someFunction(a, b, c); </syntax> <syntax type="C++"> int d, e; </syntax>

Semicolons followed by a space character, if there is more on a line

<syntax type="C++"> for (int a = 0; b++; c < d) </syntax> <syntax type="C++"> doSomething(e); doSomething(f); // This is probably bad style anyway </syntax>

Semicolons preceded by a space character, if it ends an empty loop body

It should also contain a comment to make it clear that the loop is intentionally empty. <syntax type="C++"> while (i < length - 1 && array[++i] != item) ; // Look for index of item with an empty loop </syntax> The following syntax is also acceptable: <syntax type="C++"> while (i < length - 1 && array[++i] != item) ; //this loop is intentionally empty </syntax>

When declaring class inheritance and in a ? construct, colons should be surrounded by white space

<syntax type="C++"> class BusWheel : public RubberInflatable { </syntax> <syntax type="C++"> (isNight) ? colorMeDark() : colorMeBright(); </syntax>

Indentation level is not increased after namespace clause

<syntax type="C++"> namespace Scumm {

byte Actor::kInvalidBox = 0;

void Actor::initActorClass(ScummEngine *scumm) {

   _vm = scumm;

}

} // End of namespace Scumm </syntax>

Array delete operator has no whitespace before [] <syntax type="C++"> delete[] foo; </syntax>

Template definitions

No whitespace between template keyword and < <syntax type="C++"> template<typename foo> void myFunc(foo arg) {

   // ...

} </syntax>

Operator overloading

Operator keyword is NOT separated from the name, except for type conversion operators where it is required. <syntax type="C++"> struct Foo {

   void operator()() {
       // ...
   }
   operator bool() {
       return true;
   }

}; </syntax>

Pointers and casts

No whitespace after a cast; and in a pointer, we write a whitespace before the start but not after it. <syntax type="C++"> const char *ptr = (const char *)foobar; </syntax>

References

We use the same rule for references as we do for pointers: use a whitespace before the "&" but not after it. <syntax type="C++"> int i = 0; int &ref = i; </syntax>

Switch / Case constructs

<syntax type="C++"> switch (cmd) { case kSomeCmd:

   someCmd();
   // Fall Through intended

case kSomeVerySimilarCmd:

   someMoreCmd();
   break;

case kSaveCmd:

   save();
   break;

case kLoadCmd: case kPlayCmd:

   close();
   break;

default:

   Dialog::handleCommand(sender, cmd, data);

} </syntax>

  • Note comment on whether fall through is intentional.

Naming

Constants

Basically, you have two choices (this has historical reasons, and we probably should try to unify this one day):

  1. Camel case with prefix 'k':
     kSomeKludgyConstantName 
  2. All upper case, with words separated by underscores (no leading/trailing underscores):
    SOME_KLUDGY_CONSTANT_NAME

(As a side remark, we recommend avoiding #define for creating constants, because these can lead to weird and difficult to track down conflicts. Instead use enums or the const keyword.)

Type names

Camel case starting with upper case.

<syntax type="C++"> class MyClass { /* ... */ }; struct MyStruct { /* ... */ }; typedef int MyInt; </syntax>

Class member variables

Prefixed with '_' and in camel case (Yo! no underscore separators), starting with lowercase.

<syntax type="C++"> char *_someVariableName; </syntax>

Class methods

Camel case, starting with lowercase.

<syntax type="C++"> void thisIsMyFancyMethod(); </syntax>

Local variables

Use camel case (Yo! no underscore separators), starting with lowercase.

<syntax type="C++"> char *someVariableName; </syntax>

Note that for POD structures it is fine to use this rule too.

Global variables

In general you should avoid global variables, but if it can't be avoided, use 'g_' as prefix, camel case, and start with lowercase

<syntax type="C++"> int g_someGlobalVariable; </syntax>

Special comments

Special Keywords

The following goes slightly beyond code formatting: We use certain keywords (together with an explanatory text) to mark certain sections of our code. In particular:

  • FIXME marks code that contains hacks or bad temporary workarounds, things that really should be revised at a later point.
  • TODO marks incomplete code, or things that could be done better but are left for the future.
  • WORKAROUND marks code that works around bugs in the original game, like script bugs. Sometimes these bugs worked in the original due to bugs in the original engine, sometimes the bug was visible in the original, too. It's important that you explain here what exactly you work around, and if applicable, refer to relevant tracker items!

Doxygen documentation comments

ScummVM uses the Doxygen software to generate HTML documentation for the codebase (available here).

Doxygen supports documentation blocks. These are specially-formatted comments that doxygen prints out in the generated documentation. They are similar in purpose to Java's JavaDoc or Python's docstrings.

There are many ways to mark such comments, but developers are encouraged to use the JavaDoc style:

<syntax type="C++"> /**

* Move ("warp") the mouse cursor to the specified position in virtual
* screen coordinates.
* @param x             the new x position of the mouse
* @param y             the new y position of the mouse
*/

virtual void warpMouse(int x, int y) = 0; </syntax> (See here for the docs generated from this.)

As shown in the example, documentation blocks also support several special commands such as param. Those are prefixed with either @ or \, but developers should always use @.

If you want to add a brief explanation of a variable or function after its declaration, this is the correct syntax: <syntax type="C++"> int16 x; ///< The horizontal part of the point int16 y; ///< The vertical part of the point </syntax> (See here for the docs generated from this.)

For more information, visit the official documentation:

Automatically converting code to our conventions

The following settings for Artistic Style (also known as astyle) approximate our code formatting conventions and can be used to quickly convert a big bunch of source files to our conventions. Note that you may still have to manually clean up some stuff afterwards.

indent=tab=4
brackets=attach
pad-oper
pad-header
align-pointer=name
unpad-paren
indent-preprocessor
convert-tabs