From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-176678-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 25469 invoked by alias); 13 Feb 2006 22:56:30 -0000
Received: (qmail 25453 invoked by uid 48); 13 Feb 2006 22:56:27 -0000
Date: Mon, 13 Feb 2006 22:56:00 -0000
Subject: [Bug c++/26269]  New: [3.4/4.0/4.1/4.2 regression] Declaring a variable too late yields bogus error message
X-Bugzilla-Reason: CC
Message-ID: <bug-26269-1771@http.gcc.gnu.org/bugzilla/>
Reply-To: gcc-bugzilla@gcc.gnu.org
To: gcc-bugs@gcc.gnu.org
From: "reichelt at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org>
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
X-SW-Source: 2006-02/txt/msg01424.txt.bz2
List-Id: <gcc-bugs.sourceware.org>

Compiling the invalid code snippet

==================
void foo()
{
    i;
    int i;
}
==================

yields the following error message (since GCC 3.4.5):

bug.cc: In function 'void foo()':
bug.cc:3: error: 'i' was not declared in this scope
bug.cc:4: error: redeclaration of 'int i'
bug.cc:3: error: '<typeprefixerror>i' previously declared here

The error about the redeclaraion is bogus here.
It is issued because we created a dummy declaration of 'i' to
suppress further error messages about 'i' being not declared.

Since the dummy declaration contains an error_mark_node as TREE_TYPE
(which btw. causes the <typeprefixerror> part) we should be able to
detect this case and skip issuing the redeclaration error message.
In fact we do attempt something like this in duplicate_decls:

  /* If either the type of the new decl or the type of the old decl is an
     error_mark_node, then that implies that we have already issued an
     error (earlier) for some bogus type specification, and in that case,
     it is rather pointless to harass the user with yet more error message
     about the same declaration, so just pretend the types match here.  */
  if (TREE_TYPE (newdecl) == error_mark_node
      || TREE_TYPE (olddecl) == error_mark_node)
    types_match = 1;

We indeed have TREE_TYPE (olddecl) == error_mark_node in the testcase
above, but apparently just setting types_match = 1 is not enough.

Is it possible to get rid of olddecl altogether and replace it with
newdecl?


-- 
           Summary: [3.4/4.0/4.1/4.2 regression] Declaring a variable too
                    late yields bogus error message
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Keywords: diagnostic, monitored
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: reichelt at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26269