Question

After programming in C for several years, I realized that I have been ignoring the C convention of returning zero from a function to indicate success. The convention seems semantically wrong to me, as zero is of course false. The problem is I like to name functions like is_valid_foobar(), and in order to accomodate the convention of ‘false means success’, I would have to be more vague… that is instead of:

if ( ! is_valid_foobar() ) {     return (error); } 

Other programmers write:

if ( validate_foobar() ) {      return (error); } 

And my implementation looks like:

int is_valid_foobar (int foobar ) {      if ( foobar < MAX_ALLOWED ) {           return TRUE;       }       return FALSE; } 

I haven’t actually caught any flak for this in code reviews. So I’m thinking it’s not such a terrible habit, but it is ‘unconventional’. I’m curious what people think.

I’m very careful about the choices I make for function and variable names, and a typical review comment is ‘the code is really clear’, and further it doesn’t bother me at all to have to type an extra ! at the front of the function call. But what sayest thou, oh mighty ones of S.O?

Answer 1

I’d say both are correct, for different purposes:

If you’re performing a simple go/no-go validation, e.g. is_numeric(), then true and false work nicely.

For something more elaborate, the 0==success paradigm is helpful in that it allows more than one error condition to be returned.

In this case, the caller can simply test against 0, or examine the non-0 returns for a more-specific explanation of the failure. e.g. a file open call can fail due to non-existence, insufficient permissions, etc.