Question

I developed my own Matrix class. Constructor reads a matrix from file. Matrix has free cells and “walls”. Also constructor reads start and finish points for Breadth first search (to find the shortest way from Start_point to Finish_Point).
Here is code of header:

//MyMatrix.h file

#ifndef __MYMATRIX_H__
#define __MYMATRIX_H__

#include <tchar.h>
#include <iostream>
#include <deque>

//using namespace std;
#define MAX_MATRIX_SIZE 1000

#define FREE_CELL_SIGNIFICATION '0'
#define BALL_SIGNIFICATION 'B'
#define UP_SIGNIFICATION 'U'
#define DOWN_SIGNIFICATION 'D'
#define LEFT_SIGNIFICATION 'L'
#define RIGHT_SIGNIFICATION 'R'
#define START_POINT_SIGNIFICATION 'S'
#define FINISH_POINT_SIGNIFICATION 'F'

typedef std::pair<int,int> Field_Point_Type;

//#define IS_RIGHT_NEIGHBOUR_REACHABLE(Current_Point) (((Current_Point.second+1) <= Column_Count)&&((Matrix_Field[Current_Point.first][Current_Point.second+1]==FREE_CELL_SIGNIFICATION)||(Matrix_Field[Current_Point.first][Current_Point.second+1]==FINISH_POINT_SIGNIFICATION))) ? true : false;



class Matrix {
    private:
        int Column_Count; //Cols
        int Row_Count;//Rows
        char** Matrix_Field;
        Field_Point_Type Start_Point;
        Field_Point_Type Finish_Point;
        bool Matrix_Is_Correct;
    public:
        Matrix(_TCHAR* Input_File_Name);
        int Breadth_first_search(unsigned int Start_X,unsigned int Start_Y,unsigned int Finish_X,unsigned int Finish_Y);
        ~Matrix();
        inline int IS_RIGHT_NEIGHBOUR_REACHABLE(Field_Point_Type Current_Point);
};

//MyMatrix.cpp file

...

inline int Matrix::IS_RIGHT_NEIGHBOUR_REACHABLE(Field_Point_Type Current_Point)
{      
    return  (((Current_Point.second+1) <= Column_Count)&&((Matrix_Field[Current_Point.first][Current_Point.second+1]==FREE_CELL_SIGNIFICATION)||(Matrix_Field[Current_Point.first][Current_Point.second+1]==FINISH_POINT_SIGNIFICATION))) ? true : false;
}

...

I’d like to define are the neighbour cells free for the next step of algorithm.
Of course I can use such code for this:

if (((Current_Point.second+1) <= Column_Count)&&((Matrix_Field[Current_Point.first][Current_Point.second+1]==FREE_CELL_SIGNIFICATION)||(Matrix_Field[Current_Point.first][Current_Point.second+1]==FINISH_POINT_SIGNIFICATION))) {
    //Adding of right cell to deque...
    ... 
}

but it looks ugly. I am going to add such checks for left, up and down cells.
I’d like to implement inline functions (like this: inline int IS_RIGHT_NEIGHBOUR_REACHABLE(Field_Point_Type Current_Point);).

if (IS_RIGHT_NEIGHBOUR_REACHABLE(Current_Point)) {
    //Adding of right cell to deque...
    ... 
}

It looks much better!
But I haven’t use such definition of inline function before and discovered it accidentally.
Is it good programming style?
Is it better to develop simple int IS_RIGHT_NEIGHBOUR_REACHABLE(Field_Point_Type Current_Point); method within my class?
Is it better to leave such check:

if (((Current_Point.second+1) <= Column_Count)&&((Matrix_Field[Current_Point.first][Current_Point.second+1]==FREE_CELL_SIGNIFICATION)||(Matrix_Field[Current_Point.first][Current_Point.second+1]==FINISH_POINT_SIGNIFICATION))) {
    //Adding of right cell to deque...
    ... 
}

Answer 1

I don’t think we have an established “good style” yet. Compilers capable of inlining functions from a separately compiled .cpp file are rather recent models of the most popular compilers.

Until a couple of years ago you had to have all inline functions in a .h file, so the compiler could see it while compiling the call. If your compiler is not the latest model, that might still be the rule.