From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-71477-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 20111 invoked by alias); 5 Apr 2003 12:22:55 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 20103 invoked from network); 5 Apr 2003 12:22:54 -0000
Received: from unknown (HELO thales.mathematik.uni-ulm.de) (134.60.66.5)
  by sources.redhat.com with SMTP; 5 Apr 2003 12:22:54 -0000
Received: (qmail 11564 invoked by uid 642); 5 Apr 2003 12:22:54 -0000
Message-ID: <20030405122254.11563.qmail@thales.mathematik.uni-ulm.de>
From: "Christian Ehrhardt" <ehrhardt@mathematik.uni-ulm.de>
Date: Sat, 05 Apr 2003 20:32:00 -0000
To: gcc@gcc.gnu.org
Subject: Variably sized data types and local functions
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.5.4i
X-SW-Source: 2003-04/txt/msg00220.txt.bz2


PR 8081 is basically about this piece of code (gives an ICE with all
versions of gcc):

void f (int size)
{
        typedef struct { char val[size]; } block;
        block retframe_block () { }
        retframe_block ();
}

Now my question is: Should this be legal?

Second question: Should an actual parameter of variable size be legal, i.e.:

void f (int size)
{
        typedef struct { char val[size]; } block;
        void retframe_block (block b) { }
}

None of the variations I tried currently work, but I managed to get
at least three different ICEs.

    regards  Christian

-- 
THAT'S ALL FOLKS!