From mboxrd@z Thu Jan  1 00:00:00 1970
From: "Paul N. Hilfinger" <hilfingr@otisco.mckusick.com>
To: gdb@sources.redhat.com
Subject: Scope of typedefs
Date: Mon, 01 Oct 2001 23:07:00 -0000
Message-id: <200110020607.XAA31160@localhost.localdomain>
X-SW-Source: 2001-10/msg00016.html

The code in dwarf2read and the other readers assigns all typedef symbols
to the static symbol list (file scope).  This is not, strictly speaking, 
correct, is it?  For example, in the (admittedly unusual) program

    static int Foo;

    int f (int y) {
      typedef int Foo;
      Foo x;
      x = 3;
      return x;
    }

    main () {
      f (Foo);   /* <<< */
    }

an attempt to print Foo in GDB when stopped at the commented line gives us
an

    Attempt to use a type name as an expression

message.  I understand that GCC would have to be modified to put typedefs
in proper scope, but is there any particular objection to having GDB be ready 
if it happens?  For that matter, does anyone know why things are as they are?

P. Hilfinger