Proposal of additional check

[Alexander Trufanov]

It looks I don't familiar with CPP standards much and I run into this problem today.

x = a + b?c:d;

It's a conditional assignment mixed with other operator in one statement.
x is always c or d. a is always ignored. At least for gcc version 4.9.2 (Ubuntu 4.9.2-10ubuntu13). No warnings are given during compilation.

I think there is no check for that in cppcheck. Perhaps someone could add it?

[Alexander Mai]

Why do you think a is always ignored?
In the following example you can see the results depends on a:

#include <stdio.h>
int foo(int a, int b, int c, int d) {
   int x = a + b?c:d;
   return x;
}

int main() {
  printf("%d\n", foo(0,1,3,4));
  printf("%d\n", foo(-1,+1,3,4));
  return 0;
}

You might check out http://en.cppreference.com/w/cpp/language/operator_precedence which explains the rules of operator evaluation.

[Alexander Trufanov]

Hm, then it wasn't the ignorance of a but I wasn't familiar with order of operator precedence for ?:.
I tried your code - build it with gcc version 4.9.2 (Ubuntu 4.9.2-10ubuntu13). And got output:

3
4

Then I replaced int x = a + b?c:d; with int x = a + (b?c:d);. And got

3
2

So a+b?c:d is processed as (a+b)?c:d which I didn't expect.

I would wish to get a warning from gcc on that during compilation. It barks on every else without {}.
How do you think, is it only me or it has sense to add such warning to cppcheck?

[Daniel Marjamäki]

Cppcheck should warn about this. Have you tried using --enable=style? Then maybe something in your code confuses Cppcheck, then we want to fix it.

$ cat 123.c 
int f(int a, int b) {
    int ret = a + b ? 123 : 456;
    return ret;
}
$ ./cppcheck --enable=style 123.c 
Checking 123.c...
[123.c:2]: (style) Clarify calculation precedence for '+' and '?'.