Preprocessor errors?

[Sensei]

Hi!

I'm getting some weird results from a (probably) valid C expression
with some #define... I'm really puzzled.

A sample code is this (just the variables and defines, not actual code):

#define themacro(p, i) (p)[(i)]

int *ptr, *anotherptr, counter;

themacro(ptr, counter)=anotherptr=MEMORY_ALLOCATION(...);

Where memory_allocation is a function which mallocs and reallocs
memory. I get a weird null pointer somewhere, and just writing:

anotherptr=MEMORY_ALLOCATION(...);
themacro(ptr, counter)=anotherptr;

solves the problem. Am I doing something out of standard?

 

[Artie Gold]

Hi!

I'm getting some weird results from a (probably) valid C expression with
some #define... I'm really puzzled.

A sample code is this (just the variables and defines, not actual code):

#define themacro(p, i) (p)[(i)]

int *ptr, *anotherptr, counter;

themacro(ptr, counter)=anotherptr=MEMORY_ALLOCATION(...);

Where memory_allocation is a function which mallocs and reallocs memory.
I get a weird null pointer somewhere, and just writing:

anotherptr=MEMORY_ALLOCATION(...);
themacro(ptr, counter)=anotherptr;

solves the problem. Am I doing something out of standard?

Show a compilable snippet that exhibits the problem. *Real code*. Cut
and paste.

HTH,
--ag

 

[Christopher Benson-Manica]

themacro(ptr, counter)=anotherptr=MEMORY_ALLOCATION(...);

anotherptr=MEMORY_ALLOCATION(...);
themacro(ptr, counter)=anotherptr;

solves the problem. Am I doing something out of standard?

Possibly; I'm not much of a guru to tell you. If you haven't
already, however, I would investigate your platform's documentation
for how to invoke only the preprocessor on your code (gcc -E will do
it, for example) to make sure the preprocessed line looks the way you
expect it to.

 

[Mark McIntyre]

#define themacro(p, i) (p)[(i)]

by convention, user-defined macros are UPPERCASE so they can be
identified easier.

int *ptr, *anotherptr, counter;

themacro(ptr, counter)=anotherptr=MEMORY_ALLOCATION(...);

themacro() will create an object (*ptr)[(counter)] which has type int.
This isn't compatible with anotherptr which is of type int*.

Where memory_allocation is a function which mallocs and reallocs
memory. I get a weird null pointer somewhere, and just writing:

anotherptr=MEMORY_ALLOCATION(...);
themacro(ptr, counter)=anotherptr;

solves the problem.

this is just as bad - if you don't get a warning here, you should turn
up warninglevels in your compiler.

Am I doing something out of standard?

Remember how macro substitution works - its literal text replacement.
So you can write down what themacro() will do.