开发者

Best practice: ordering of public/protected/private within the class definition?

I am starting a new project from the ground u开发者_运维知识库p and want it to be clean / have good coding standards. In what order do the seasoned developers on here like to lay things out within a class?

A : 1) public methods 2) private methods 3) public vars 4) private vars

B : 1) public vars 2) private vars 3) public methods 4) private methods

C : 1) public vars 2) public methods 3) private methods 4)private vars

I generally like to put public static vars at the top, but then would a public static method be listed ahead of your constructor, or should the constructor always be listed first? That sort of thing...

I know it's finnicky but I just wondered: what are best practices for this?

PS: no I don't use Cc#. I know. I'm a luddite.


In Clean Code, Robert C. Martin advises coders to always put member variables at the top of the class (constants first, then private members) and methods should be ordered in such a way so that they read like a story that doesn't cause the reader to need to jump around the code too much. This is a more sensible way to organize code rather than by access modifier.


The best practice is to be consistent.

Personally, I prefer putting public methods first, followed by protected methods, following by private methods. Member data should in general always be private or protected, unless you have a good reason for it not to be so.

My rationale for putting public methods at the top is that it defines the interface for your class, so anyone perusing your header file should be able to see this information immediately.

In general, private and protected members are less important to most people looking at the header file, unless they are considering modifying the internals of the class. Keeping them "out of the way" ensures this information is maintained only on a need to know basis, one of the more important aspects of encapsulation.


Personally I like to have public at top, protected and then private. The reason for this is that when somebody cracks open the header he/she sees what he/she can access first, then more details as he/she scrolls down.

One should not have to look at the implementation details of a class in order to use it, then the class design is not done well.


I think I have a different philosophy on this than most. I prefer to group related items together. I can't stand having to jump around to work with a class. The code should flow and using a rather artificial ordering based on accessibility (public, private, protected etc. ) or instance versus static or member versus property versus function doesn't help keep a nice flow. So if I nave a public method Method that is implemented by private helper methods HelperMethodA, HelperMethodB etc. then rather than have these method far apart from each other in the file, I will keep them close to each other. Similarly, if i have an instance method that is implemented by a static method, I will group these together too.

So my classes often look like this:

class MyClass {
    public string Method(int a) {
        return HelperMethodA(a) + HelperMethodB(this.SomeStringMember);
    }

    string HelperMethodA(int a) { // returns some string }

    string HelperMethodB(string s) { // returns some string }

    public bool Equals(MyClass other) { return MyClass.Equals(this, other); }

    public static bool Equals(MyClass left, MyClass right) { // return some bool }

    public double SomeCalculation(double x, double y) {
        if(x < 0) throw new ArgumentOutOfRangeException("x");
        return DoSomeCalculation(x, y); 
    }

    const double aConstant;
    const double anotherConstant;
    double DoSomeCalculation(double x, double y) {
        return Math.Pow(aConstant, x) * Math.Sin(y) 
            + this.SomeDoubleMember * anotherConstant;
    }       
}


This would be my ordering

  1. Static Variables
  2. Static Methods
  3. Public Variables
  4. Protected Variables
  5. Private Variables
  6. Constructors
  7. Public Methods
  8. Protected Methods
  9. Private Methods

I use the following rules:

  • static before anything
  • variables before constructors before methods (i consider constructors to be in the category of methods)
  • public before protected before private

The idea is that you define the object (the data), before the behaviours (methods). Statics need to be separated because they aren't really part of the object, nor it's behaviour.


I used to care a lot. Over the last several years using modern IDEs pretty much everything is only 1 or 2 keystrokes away, I've let my standards relax substantially. Now, I start with statics, member variables, then constructors after that I don't worry about it much.

In C# I do let Resharper organize things automatically.


I generally agree with the public, protected, private order as well as the static data, member data, member functions order.

Though I sometimes group like members (getters & setters) I generally prefer listing members within a group ALPHABETICALLY so that they can be located more easily.

I also like lining up the data/functions vertically. I tab/space over to the right enough so that all names are aligned in the same column.


To each his own, and as Elzo says, modern IDEs have made it easier to find members and their modifiers in an easy way with colored icons in drop-down menus and such.

My take is that it is more important for the programmer to know what the class was designed for, and how it can be expected to behave.

So, if it is a Singleton, I put the semantics (static getInstance() class) first.

If it is a concrete factory, I put the getNew() function and the register / initialize functions first.

... and so on. When I say first, I mean soon after the c'tors and d'tor - since they are the default way of instantiating any class.

The functions that follow are then in:

  1. logical call-order (e.g. initialize(), preProcess(), process(), postProcess() ), or
  2. related functions together (like accessors, utilities, manipulators etc),

depending on if the class was meant primarily to be a data-store with some functions, or function provider with a few data members.


Some editors, like Eclipse and its offspring, allow you to reorder in the outline view the the vars and the methods, alphabetically or as in page.


The sequence of public followed by protected and private is more readable to me, It's better to describe the class logic in comments at top of the header file simply and function call orders to understand what a class dose and algorithms used inside.

I am using Qt c++ for a while and see some new sort of keywords like signal and slot I prefer to keep ordering like above and share my idea with you here.

#ifndef TEMPLATE_H
#define TEMPLATE_H


class ClassName
{
    Q_OBJECT
    Q_PROPERTY(qreal startValue READ startValue WRITE setStartValue)
    Q_ENUMS(MyEnum)

public:

    enum MyEnum {
        Hello = 0x0,
        World = 0x1
    };

    // constructors

    explicit ClassName(QObject *parent = Q_NULLPTR);
    ~ClassName();

    // getter and setters of member variables

    // public functions (normal & virtual) -> orderby logic

public slots:

signals:

protected:

    // protected functions it's rule followed like public functions


private slots:

private:

    // methods

    // members

};

#endif // TEMPLATE_H
0

上一篇:

下一篇:

精彩评论

暂无评论...
验证码 换一张
取 消

最新问答

问答排行榜