RE: nonstandard assumptions in the code

[Hallvard B Furuseth <h.b.furuseth@usit.uio.no>]

Howard Chu writes:

>>>> * can cast function pointers to 'void *' and back
>>>>   (at least the threading stuff)
> 
> K&R 2nd Edition (ANSI C) p.120 "Any pointer can be cast to void * and back
> again without loss of information."

This does not apply to function pointers.  There have been plenty of
threads about this on comp.std.c and comp.lang.c.  Some cite real-world
examples.

Formal reason: ANSI 3.2.2.3 says 'A pointer to any INCOMPLETE OR OBJECT
TYPE may be converted to a pointer to void and back again; the result
shall compare equal to the original pointer.'
However, a function is not an object.

>> Another non-ANSI thing: LBER_OPT_ON and LDAP_OPT_ON are #defined as
>> ((void*)1).  There is no guarantee that this is != NULL (the *_OPT_OFF
>> options).  I'll install this fix unless someone protests:
> 
> K&R 2nd Edition p.102 - NULL is zero, by definition. Also stated again
> in section 6.2.2.3 of the C90 spec.

Not quite.  NULL is a null pointer constant, and a null pointer constant
is _spelled_ 0 or (void*)0.  But that doesn't mean that some other
integer cast to void* can't compare equal to NULL.  The result of
casting integers (other than the constant 0) to pointers is
implementation-defined.

-- 
Hallvard