Assert - segmentation fault

Question

I've got an issue i can't handle so i've thought maybe you can help me. Basically i have a function that receives a char* as a paramater and does some stuff to it(i've checked those steps/functions and they work just fine).If the char* given in the function is ""(i guess NULL), i receive seg fault with assert. Here is the code:

char *computeRNA(char *s)
{
    if (s != NULL && s!= "")
    {
        Lista* l = to_list(s);
        int i;
        l = l->next;
        for(i = 0; i<= lungime(l); ++i)
        {
            if(l->info == 'T')
                l->info = 'U';
            l = l->next;
        }
        char *rna = to_pointer(l);
        return rna;

    }
        return NULL;
}

And here is the assert:

 char *s;
   s = computeRNA("");
    ASSERT(!strcmp(s, ""), "computeRNA-01");
    free(s);

This is a school homework so i can not change assert's code , only the function.Thanks in advance !

and while the first in @BLUEPIXY's example is a stylistic change, the second certainly isn't. — Antti Haapala -- Слава Україні, Mar 11 '16 at 06:58
You're supposed to return an empty string when given an empty string! — Antti Haapala -- Слава Україні, Mar 11 '16 at 06:59
Applying the `=` operator to a C-"string" (which in fact is a `char`-array) does not what you seem to expect. — alk, Mar 11 '16 at 07:00
[How do I properly compare strings in C?](http://stackoverflow.com/questions/8004237/how-do-i-properly-compare-strings-in-c) — Lundin, Mar 11 '16 at 07:20
Also what is the `free(s)` doing there? There is no dynamic memory. — Lundin, Mar 11 '16 at 07:21
@Lundin: `free(s)` is probably part of a test, implying `computeRNA` must return a pointer returned from `malloc`. — Michael Foukarakis, Mar 11 '16 at 07:22

Sourav Ghosh · Answer 1 · 2016-03-11T07:16:59.583

0

You cannot pass a NULL pointer to strcmp() as argument. Before the ASSERT statement, just check for non-NULL value of s. Something like

if (s) {
  ASSERT(!strcmp(s, ""), "computeRNA-01");
 }

should do the job.

EDIT:

In case changing the function using the assertion is also not possible, you can modify the computeRNA() function to return an empty string, instead of NULL, like

char * empty_string = calloc(1,1);
return empty_string;

which can later be passed to free() in the caller.

edited Mar 11 '16 at 07:16

answered Mar 11 '16 at 06:56

Sourav Ghosh

133,132
16
183
261

well OP said that asserts are not to be changed, they're checking that the algorithm works as required... – Antti Haapala -- Слава Україні Mar 11 '16 at 07:00
@AnttiHaapala I'm not sure if I'm suggesting change to the `ASSERT`, rather the function using the assertion. I thought that was allowed. – Sourav Ghosh Mar 11 '16 at 07:01
case of `return "";` : can't free it. – BLUEPIXY Mar 11 '16 at 07:13
@BLUEPIXY OK, right... Let me correct my answer then. – Sourav Ghosh Mar 11 '16 at 07:15
It is the worst of the specification request. (I say to school) (; ・ ;)sad (I alread upvoted) – BLUEPIXY Mar 11 '16 at 07:23

Antti Haapala -- Слава Україні · Accepted Answer · 2016-03-11T16:05:03.883

This ASSERT that you weren't supposed to be changing, tests that if an empty string is given to computeRNA it should also return an empty string:

s = computeRNA("");
ASSERT(!strcmp(s, ""), "computeRNA-01");

Your code returns null pointer. Also comparing strings needs to be done using strcmp, s == "" would only compare pointer values that wouldn't be of use here.

Thus I'd start the function with

// null pointer was given
if (!s) {
    return calloc(1, 1);
}

// an empty string was given (the first character pointed to b s is the string terminator.)
// to compare explicitly, you need to do `strcmp(s, "")` == 0. `s == ""` is wrong.
if (!*s) {
    return "";
}

Assert - segmentation fault

2 Answers2