[Bug c/20519] [4.0/4.1 regression] completed type not selected properly with typeof

[joseph at codesourcery dot com]

------- Additional Comments From joseph at codesourcery dot com  2005-03-17 16:25 -------
Subject: Re:  [4.0/4.1 regression] completed type not selected
 properly with typeof

On Thu, 17 Mar 2005, rth at gcc dot gnu dot org wrote:

> Array dimension data gets stripped here:
> 
> c-decl.c:grokdeclarator
> 3863      if (!flag_gen_aux_info && (TYPE_QUALS (element_type)))
> 3864        type = TYPE_MAIN_VARIANT (type);

Looks like complete_array_type is broken in that it should complete the 
unqualified type then build the array of qualified type with the proper 
TYPE_MAIN_VARIANT.  Because my audit was auditing all uses of 
build_array_type and TYPE_MAIN_VARIANT and complete_array_type is 
modifying types directly, this didn't get caught.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20519