C and C++ Programming
Avoiding Assignment/Equality Bugs
If compairing a variable to a constant, the constant should be the lvalue (on the left).
Fo example, this erroneous code will compile without errors (good compilers will warn you):
while (foo = NULL)
{ ... }
The pointer foo will be set to NULL and the loop will never be entered. On the other hand, trying to build the following will always generate an error:
while (NULL = foo)
{ ... }
test.c: In function `main': test.c:7: error: invalid lvalue in assignment
Using Bitflags
Bitflags are a method of storing multiple flags, which are not mutucally exclusive, in one variable.
You’ve probably seen them before in system or library calls. Each flag is a bit position which can be set on or off. You then have a bunch of bitmasks declared with #define for each bit position so you can easily manipulate it:
#define LOG_ERRORS 1 // 2^0, bit 0 #define LOG_WARNINGS 2 // 2^1, bit 1 #define LOG_NOTICES 4 // 2^2, bit 2 #define LOG_INCOMING 8 // 2^3, bit 3 #define LOG_OUTGOING 16 // 2^4, bit 4 #define LOG_LOOPBACK 32 // and so on... unsigned flags; // WARNING: assigning a value will clobber any other flags, so you // should generally only use the = operator when initialising vars. // initialising the flags flags = LOG_ERRORS; // sets to 1 i.e. bit 0 //initialising to multiple values with OR (|) flags = LOG_ERRORS | LOG_WARNINGS | LOG_INCOMING; // sets to 1 + 2 + 8 i.e. bits 0, 1 and 3 // setting one flag on, leaving the rest untouched // OR bitmask with the current value flags |= LOG_INCOMING; // testing for a flag // AND with the bitmask before testing with == if ((flags & LOG_WARNINGS) == LOG_WARNINGS) ... // testing for multiple flags // as above, OR the bitmasks if ((flags & (LOG_INCOMING | LOG_OUTGOING)) == (LOG_INCOMING | LOG_OUTGOING)) ... // removing a flag, leaving the rest untouched // AND with the inverse (NOT) of the bitmask flags &= ~LOG_OUTGOING; // toggling a flag, leaving the rest untouched flags ^= LOG_LOOPBACK;
A variable that’s really a function
Stolen from the FreeBSD errno manpage:
extern int * __error(); #define errno (* __error())
ANSI codes and colours
Before you start, make sure you’re checking terminal types or at least providing a monochrome option so people who use terms that don’t support this don’t suffer excape characters all over their text.
The codes themselves have a starting sequence, (Escape-[ or \033[) then a set of numbers seperated by semicolons which may contain a text attribute normal, bold, underlined, blinking…) and foreground and background colours. The code is closed with an m.
Basic ANSI Properties:
Enable Disable Foreground Background
0 normal 30 black 40 black
1 bold 22 unbold 31 red 41 red
4 underline 24 ununderline 32 green 42 green
5 blink 25 unblink 33 yellow 43 yellow
7 inverse 27 uninverse 34 blue 44 blue
9 strike 29 unstrike 35 magenta 45 magent
36 cyan 46 cyan
37 white 47 white
39 default 49 default
Attributes are not as widely supported as simple colours. Often in an X environment the user has chosen a bold font and bold will have no effect; underlined, blinking and inverse text sometimes fail for no particular reason. The codes for cancelling an attribute (unbold etc) sometimes do something completely different. Strikethrough is “standard” but I’ve never seen it actually work.
For these reasons you’re probably better off just sticking with the colours. A C example follows:
#define ANSI_NORMAL "\033[0m"
/* these are all normal text with black background (40) */
#define ANSI_RED "\033[0;31;40m"
#define ANSI_GREEN "\033[0;32;40m"
#define ANSI_YELLOW "\033[0;33;40m"
#define ANSI_BLUE "\033[0;34;40m"
#define ANSI_MAGENTA "\033[0;35;40m"
#define ANSI_CYAN "\033[0;36;40m"
#define ANSI_WHITE "\033[0;37;40m"
#define ANSI_BOLD_RED "\033[1;31;40m"
#define ANSI_UNDERLINE_RED "\033[4;31;40m"
#define ANSI_BLINK_RED "\033[5;31;40m"
printf("%sI am blue!%s\n", ANSI_BLUE, ANSI_NORMAL);
Always, always close the text with the 033[0m code (normal text) or else the text after your program ends will be affected…
Polymorphic Structures in C
This can be achieved with a little creative union and struct work. The basic idea is to have a structure representing a superclass, which contains a union of all the other subclasses.
You will need another entity (preferably an enumeration) in the superclass and all subclasses to determine what type of struct it is.
Example:
typedef struct{
int type;
float ypos;
float xpos;
float length;
float height;
} Shape_Triangle
typedef struct{
int type;
float ypos;
float xpos;
float length;
} Shape_Square
typedef struct{
int type;
float ypos;
float xpos;
float radius;
} Shape_Circle
typedef union{
int type;
Shape_Square square;
Shape_Triangle triangle;
Shape_Circle circle;
} Shape;
...
Shape s = getShape();
switch(s.type)
{
case SHAPE_SQUARE:
s.Shape_Square.length=3;
break;
case SHAPE_TRIANGLE:
s.Shape_Triangle.height=4;
break;
case SHAPE_CIRCLE:
s.Shape_Circle.radius=5;
break;
}
A drawback of this method is that you need to duplicate the members in the substructure, and you must make sure the type variables are all in the same physical position (first is best) in the struct/union. Alternatively, use a struct for the superclass and subclass:
typedef struct{
float length;
float height;
} Shape_Triangle
typedef struct{
float length;
} Shape_Square
typedef struct{
float radius;
} Shape_Circle
typedef union{
Shape_Square square;
Shape_Triangle triangle;
Shape_Circle circle;
} Shape_Union;
typedef struct{
int type;
float xpos;
float ypos;
Shape_Union subshape;
}
...
Shape s = getShape();
switch(s.type)
{
case SHAPE_SQUARE:
s.subshape.Shape_Square.length=3;
break;
case SHAPE_TRIANGLE:
s.subshape.Shape_Triangle.height=4;
break;
case SHAPE_CIRCLE:
s.subshape.Shape_Circle.radius=5;
break;
}
This approach requires extra typing “through” the union.