CODEWITHSUNDEEP
cvisual-studiopointersstruct
K.H.A.J.A.S
K.H.A.J.A.S

Reputation: 516

Why is this initialization to a static pointer not valid?

I am trying to write a macro to initialize fields of a complex structure. In this structure, some member is a pointer which I want to initialize sometimes with a real address and sometimes to NULL. The structure can be simplified as follows:

typedef struct {
int * p;
} MYSTRUCT;

My first pass at the macro was:

#define INIT_STRUCT(x) {&(x)}

and the usage would be:

static int foo;
static MYSTRUCT struct1 = INIT_STRUCT(foo);

This works well when I initialize to a real pointer, but the following does not work:

static MYSTRUCT struct2 = INIT_STRUCT(NULL);

because it is developed as {&(NULL)} and the compiler (justifiably) flags having & on a constant as an error. I know that I could choose to have the & as part of the argument and not part of the macro body, but this is not convenient in my case, and also I cannot accept that the compiler wins, so I tried to be smarter with the following second version of the macro:

static void * NOTHING;
#define INIT_STRUCT(x) {&NOTHING == &(x) ? NULL : &(x)}

with usage

MYSTRUCT struct1 = INIT_STRUCT(foo);
MYSTRUCT struct2 = INIT_STRUCT(NOTHING);

However the compiler protests that "initializer is not a constant". If I replace &NOTHING == &(x) by 1 == 2, there is no error, so the format of a condition is not the problem. On the other hand, &NOTHING and &(foo) are valid as initializers by themselves.

Is the compiler correct in invalidating my "smarter" macro? Is there any other solution, except writing the & as part of the argument for foo?

The compiler is Microsoft Visual Studio C 2010 Ultimate.

Upvotes: 0

Views: 118

Answers (1)

Rishikesh Raje
Rishikesh Raje

Reputation: 8614

You can simplify and use this.

#define INIT_STRUCT(str, x) {str -> p = x}

This is used as 

INIT_STRUCT(struct1, foo);
INIT_STRUCT(struct2, NULL);

Upvotes: 1

Related Questions