Thread: Error Handling Advice Wanted

Say I have:

Code:

int sl_find(StringList *pSl, const char *str)
{
    if(NULL == pSl || NULL == str)
    {
        return FAIL;
    }
    size_t i;
    for(i = 0; i < pSl->size; i++)
    {
        if(strcmp(pSl->data[i], str) == 0) return i;
    }
    return -2;
}

Where #define FAIL -1

Its purpose is to locate a substring in the stringlist. If we find, we return the position. If we don't find, we return -1. What would you return in an error case such as NULL input ptrs? I don't want to return -1 becausae that would be ambiguous with "not found" but -2 is also not super conventional. I know an alternative is to use an output param but that feels heavy and a bit untuitive to either store the index or an error code.
Just high level, how do you reason about these types of things?
Not necessarily specific to this exact code. I've looked at for example the standard string.h functions and the return codes don't seem to be entirely consistent throughout.

IMO, it would be pretty pointless to constantly check for NULL pointers and return some kind of code to the caller in every function that receives a pointer. The caller can ensure his inputs are not NULL in the first place. If they screw up with that who's to say they won't screw up in testing for your special value?

It's more usual to do something like this so that you can have the NULL check in place for testing but remove it later on.

Code:

int sl_find(StringList *pSl, const char *str)
{
    assert(pSl && str);
    for (size_t i = 0; i < pSl->size; i++)
        if (strcmp(pSl->data[i], str) == 0)
            return i;
    return -1; // don't name this; -1 is good enough
}

Can you demonstrate a couple of those inconsistencies that you found in the standard library implementation that you studied? What exactly do you mean?