MaxScale/Development/coding-style-and-guidelines.md
Johan Wikman 81297fb919 Update naming convention for structs
The earlier guideline was just a "formalization" of existing
conventions.
2018-08-02 14:16:53 +03:00

6.7 KiB

The purpose of a coding style and accompanying guidelines is to make the reading and comprehension of the code easier. Reading prose would also be much harder, should the punctuation rules differ from chapter and paragraph to the next.

Coding Style

astyle

MaxScale comes with a Astyle configuration file used to format the source code. To use it, run the following in the source root.

astyle --options=astylerc <path to source>

This will format the source file according to the MaxScale coding style.

Consistency

Be consistent. In a particular context, everything should style wise look the same.

In Rome do as the romans do.

If something is done in a particular way in some context, then follow that same way when doing modifications, even if it violates what is stated here. If you want to fix the style of a file, then do so in one separate change; not as part of other modifications.

General

  • Only spaces, no tabs.
  • Indentation depth 4 spaces.
  • No trailing white space.
  • Maximum line length 110 characters.

Indentation Style

We follow the Allman indentation style.

Braces are always used and the brace associated with a control statement is placed on the next line, indented to the same level as the control statement. Statements within the braces are indented to the next level.

if (something)
{
    do_this();
}
else
{
    do_that();
}

Punctuation

Keywords are followed by a space.

if (something)
{
   ...;
}

while (something)
{
   ...;
}

Exceptions are sizeof, typeof and alignof.

Function name and opening parenthesis are not separated by a space.

some_function();

Operators

Use one space around (on each side of) most binary and ternary operators, such as any of these:

=  +  -  <  >  *  /  %  |  &  ^  <=  >=  ==  !=  ?  :

but no space after unary operators:

&  *  +  -  ~  !  sizeof  typeof  alignof  __attribute__  defined

no space before the postfix increment & decrement unary operators:

++  --

no space after the prefix increment & decrement unary operators:

++  --

and no space around the . and -> structure member operators.

Comma is always followed by a space.

x = some_function(a, b, c + d);

Opening parenthesis and square bracket are not followed by a space, and closing parenthesis and square bracket are not preceded by a space.

int len = strlen(name);

a = b[5];

Naming

enums

enum gwbuf_type { ... };
typedef enum { ... } gwbuf_type_t;
typedef enum gwbuf_type { ... } gwbuf_type_t;

structs

A struct must be a POD type. It must have no non-POD members and it must not have any member functions. Whether a struct has been declared in a C or C++ header file, it must be declared as if it would be used from C.

In a C header, a struct is declared as:

typedef struct SOME_TYPE { ... } SOME_TYPE;

With this arrangement it is possible to refer to the type using SOME_TYPE or struct SOME_TYPE from both C and C++ code.

In a C++ header, a struct is declared without the typedef.

struct SOME_TYPE;

functions

Small caps and words separated by underscores

void server_set_status(SERVER *, int);

with the exception of names of function pointers in plugin interfaces

typedef struct filter_object {
    FILTER *(*createInstance)(char **options, FILTER_PARAMETER **);
    ...
} FILTER_OBJECT;

that are camelCase.

C++ naming

namespaces

Small caps and words separated by underscores

namespace maxscale
{

class ...

}
...
namespace xyz_filter
{
...
}

Note that symbols within a namespace are not indented. Note also that the namespace maxscale can only be used by classes belonging to the MaxScale core. An exception is when a template in the MaxScale namespace is specialized for a non MaxScale core class.

classes

To distinguish them from plain-old-data, aka structs, class names use camel-case.

class CacheFilterSession
{
public:
    int some_member_function();
    ...
};

Note that the naming of member functions follows the same conventions as the naming of free functions, that is, snake-case is used.

member variables

Member variables are prefixed with m_ and static member variables are prefixed with s_.

class MyClass
{
    ...
private:
    int m_size;
    static int s_max_size;
};

In general, a class should have no public member variables but all access should be via public functions that can be inline.

variable prefixes

The following prefixes are "standardized" and can be used when applied consistently:

int *pAge;                        // p, for pointers.
shared_ptr<NiftyBuffer> sBuffer;  // s, for smart pointers
unsigned nItems;                  // n, when a variable denotes a count.

Note that there is no underscore between the prefix and the actual variable name, and that the character following the prefix is in large caps. Note also that the prefixes can be stacked.

class SomeClass
{
private:
    int* m_pIndex;
    int** m_ppIndex;
    shared_ptr<SomeClass> m_sNext;
};

Guidelines

These are suggestions.

Try to have only a single exit-point in each function.

Rationale: With multiple exit points it is very easy to overlook the release of some resource in one of those places.

Reasonable exceptions to the single exit point rule are:

  • Checking of preconditions at the beginning of the function.
 int function(int a, int b)
 {
     if (a < 0)
         return -1;

     if (b > 5)
         return -1;

     int rv;
     ...
     return rv;
}
  • Functions that are basically only big switches.
char* toString(enum Color c)
{
    switch (c)
    {
        case RED:
            return "RED";

        case GREEN:
            return "GREEN";

        default:
            return "Unknown";
   }
}

Try to keep the line length less than 80 characters.

Rationale: In typography the optimal line length for readability is considered to be around 60 characters. With 5 levels of nesting of 4 spaces each, that gives 80 characters. If it is hard to stay within that limit, consider refactoring the function.

Use parenthesis to make the precedence explicit, even if precedence rules would not require it.

if (a != b && !ready()) ...;

if ((a != b) && !ready()) ...;

Rationale: There can never be any doubt about what the intended precedence is.

Try to keep the length of each function less than what fits in one screen full (roughly 60 lines).

Rationale: That way it is possible with one glance to get a grasp of what the function does. If it's hard to keep within the limit it may be a sign that the function ought to be refactored.

Use vertical space (i.e. empty lines) to introduce paragraphs and sections.

Rationale: Makes the reading easier.