[Bug c/33076] Warning when passing a pointer to a const array to a function that expects a pointer to a non-cast one

["raeburn at raeburn dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #2 from raeburn at raeburn dot org  2007-08-15 06:15 -------
Subject: Re:   New: Warning when passing a pointer to a const array to a
function that expects a pointer to a non-cast one

On Aug 14, 2007, at 23:45, martin dot ferrari at gmail dot com wrote:
> Sorry if I'm misunderstanding something or this is a FAQ, but I  
> couldn't find
> any reference, except for a unanswered mail with the same problem  
> from 2000 at
> http://gcc.gnu.org/ml/gcc-bugs/2000-10/msg00337.html (and I don't  
> find it in
> bugzilla either). The code looks semantically correct to me, the  
> function
> declares that it won't touch the data and I'm passing a non-const  
> data. If the
> typecast is not to an array, the warning goes away.


> /home/martin/test.c:8: warning: passing argument 1 of ?foo? from  
> incompatible
> pointer type

> typedef int caca[3];
> void foo (const caca* p);
> void bar(caca *c) {
>         foo(c);
> }

I think this might be related to the bug report I filed in 31887 --  
and I'm starting to think that we might both be wrong.

Weird as it seems, my reading of the C spec, section 6.7.3 says that  
the qualifier (const, in both our cases, but it would apply to  
volatile as well) applies only to the array element type, and does  
*not* apply to the array type itself.  So this isn't like assigning a  
"char *" value to a "const char *"; you're dealing with pointers to  
two different array types (which happen to have similar but  
differently-qualified element types).

I'm not 100% sure I believe this interpretation, because it seems to  
needlessly prohibit perfectly reasonable code, and when I asked on  
comp.std.c a couple weeks back, I got some mixed feedback.  But at  
the moment, that's my reading of the spec....

Ken


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33076