Segfault using logical OR on for loop.

I've received a segmentation fault in this part of code:

Code:

---

for (; str[i] != '\0' || str[i] != ' '; i++)
{
        result[result_pos] = str[i];
        result_pos++;
}

---

which, should mean:

Code:

---

If str[i] is not equal to NULL character nor ' '; then, keep printing the chars within the word to variable result.

---

To understand what the function does, here's the whole code:

Code:

---

static char result[999];

/* char *split_word(const char *str, long n);
 * Splits a string word-by-word. Where N is the word number. */
char *split_word(const char *str, long n)
{
    size_t result_pos = 0;
    long scr;
    size_t i = 0;

    for (scr = 0; scr < n; scr++)
    {
        for (; str[i] != ' '; i++) ; /* Scrolls forward to the desired word */
        i++; /* Skip one char (the space). */
    }

    for (; str[i] != '\0' || str[i] != ' '; i++)
    {
        result[result_pos] = str[i];
        result_pos++;
    }

    return result;
}

---

I'm trying to accomplish an easier way round to split string by word than using "strtok_r()". It might sound like "re-inventing the wheel", but... i will try to make it neat later.

The code looks okay, why would it segfault? how can i fix that?

There is NOTHING that makes this false!

Code:

---

str[i] != '\0' || str[i] != ' '

---

If str[i] equals '\0' then it CAN NOT equal a space; therefore it will always be true.

Tim S.

i don't understand. :/

i thought it would do:

str[i] is not equal to '\0'; ========>
+--------------+
| Continue loop |
+--------------+
str[i] is not equal to ' '; =========>

so, if both statements are says that str[i] != '\0' nor ' '; then, run code in loop.
until, str[i] is equal to '\0' or str[i] is equal to ' '; then, stop the loop.

Quote:

---

Originally Posted by milo64

Code:

---

for (; str[i] != '\0' || str[i] != ' '; i++)
{
        result[result_pos] = str[i];
        result_pos++;
}

---

which, should mean:

Code:

---

If str[i] is not equal to NULL character nor ' '; then, keep printing the chars within the word to variable result.

---

---

That description is ambiguous and certainly does match what the code unambiguously does.

The code will keep looping forever. There are no characters for which the terminating condition is false. That means i is continually incremented and will eventually exceed 999 (so, by writing to result[i], the code trashes a random area of memory).

If str[i] is non-zero, the loop condition is always true (and the second test will not be performed). if str[i] is zero, the test str[i] != 0 will be false so the test str[i] != ' ' will be performed which - since str[i] is zero, otherwise it would not have yielded false for the first test - will always yield true.

How do you fix it? First describe what the code is intended to do, unambiguously - otherwise you can't hope to code something up and have it reliably work. Find a loop condition that can actually yield a false result, so the loop can terminate.

Quote:

---

Originally Posted by grumpy

That description is ambiguous and certainly does match what the code unambiguously does.

The code will keep looping forever. There are no characters for which the terminating condition is false. That means i is continually incremented and will eventually exceed 999 (so, by writing to result[i], the code trashes a random area of memory).

If str[i] is non-zero, the loop condition is always true (and the second test will not be performed). if str[i] is zero, the test str[i] != 0 will be false so the test str[i] != ' ' will be performed which - since str[i] is zero, otherwise it would not have yielded false for the first test - will always yield true.

How do you fix it? First describe what the code is intended to do, unambiguously - otherwise you can't hope to code something up and have it reliably work. Find a loop condition that can actually yield a false result, so the loop can terminate.

---

So, you mean, the != returns the false, so, i should find a kind of loop that would exit if the statement is false? right?

I think the point is, you claim to want to break the loop when the character is neither \0 NOR a space, but that's not the logic you've given. The logic you've given could never return false.

"Neither A nor B" would, in logical terms, be "not (A or B)".

"not (A or B)" is completely different from "(not A) or (not B)", which is what you've coded. You might consider DeMorgan's laws helpful:

!(A || B) = !A && !B
!(A && B) = !A || !B

No.

Firstly, your problem is not in the kind of loop. No loop (a for-loop, a while-loop, or a do-while-loop) will ever terminate with that loop condition.

Second, the problem has nothing to do with the != operator. The != tests can each yield true or false. Your problem is in how you are combining the two tests.

And, no, I'm not going to be more specific than that. A bit of time spent nutting out the problem will teach you a lot. I've told you what is wrong. You have to find an approach to correct that.

Anyway, i changed it to this way:

Code:

---

for (;; i++)
    {
        if (str[i] != '\0' && str[i] != ' ')
        {
            result[result_pos] = str[i];
            result_pos++;
        }
        else
            break;
    }

---

thanks all for you effort. I was a bit confused.