added string comparison mu_assert_string_eq

[spencerbug]

string comparison is a very useful feature for many tests.
This particular one also checks if the strings are null, and replaces those with "", before doing the comparison.

[soudy]

+1, would save a lot of repetition in my tests

[spencerbug]

this should be const char*

[siu]

I would like to merge this pull request but the code is not 100% safe. I think I mentioned this earlier but the comment disappeared after the rebase: when dealing with macros one has to be careful when the arguments appear in the expanded body.

With the current implementation actual and expected appear more than once in the macro. If actual or expected are expressions with side effects it will have unexpected behavior from the caller side.

A concrete example, if result = "queue.popAndReturn()" which removes and returns the element on top of a queue. The macro will be expanded to (only expanding expected):

MU__SAFE_BLOCK(
    const char* minunit_tmp_e;
    const char* minunit_tmp_r;
    minunit_assert++;
    if (!queue.popAndReturn()) {
        minunit_tmp_e = "<null pointer>";
    } else {
        minunit_tmp_e = (queue.popAndReturn());
    }
    if (!result) {
        minunit_tmp_r = "<null pointer>";
    } else {
        minunit_tmp_r = (result);
    }
    if(strcmp(minunit_tmp_e, minunit_tmp_r)) {
        snprintf(minunit_last_message, MINUNIT_MESSAGE_LEN, "%s failed:\n\t%s:%d: '%s' expected but was '%s'", __func__, __FILE__, __LINE__, minunit_tmp_e, minunit_tmp_r);
        minunit_status = 1;
        return;
    } else {
        printf(".");
    }
)

I.e. it will potentially invoke popAndReturn twice and the logic would be wrong.

I suggest to assign const char* minunit_tmp_e = expected; and const char* minunit_tmp_r = result; at the beginning of the macro and use the temporary variables for all the comparisons.

[siu]

Looks good now, thanks for the contribution.

[R-Gerard]

Documentation in the README was omitted.