Re: [RFC] Cleaning and extening the diagnostics machinery

["Joseph S. Myers" <jsm28 at cam dot ac dot uk>]

On 31 May 2001, Gabriel Dos Reis wrote:

> A warning for an unimplemented fonctionality?  What usage do you see
> for it?

There are at least a couple of uses of this in the C front end:

c-parse.in (typename):

                  /* We don't yet support attributes here.  */
                  if (attrs != NULL_TREE)
                    warning ("attributes on type name ignored");

(There are many places where attributes ought to work, but don't and have 
funny effects.  The normal practice of GCC on attributes it doesn't 
understand or handle is to ignore them with a warning - in this case, the 
attributes may be valid, but this place for them isn't implemented.)

c-decl.c (build_array_declarator):

  if (vla_unspec_p)
    warning ("GCC does not yet properly implement `[*]' array declarators");

(Unless you rely on [*] representing an VLA type that is complete but of 
indeterminate size, rather than an incomplete type, or on the proper 
constraint checking, this C99 syntax can be used.)

-- 
Joseph S. Myers
jsm28@cam.ac.uk