From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-107955-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 12822 invoked by alias); 11 Jan 2005 16:04:58 -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 12223 invoked from network); 11 Jan 2005 16:03:32 -0000
Received: from unknown (HELO mxfep02.bredband.com) (195.54.107.73)
  by sourceware.org with SMTP; 11 Jan 2005 16:03:32 -0000
Received: from wigwam ([213.112.182.104] [213.112.182.104])
          by mxfep02.bredband.com with ESMTP
          id <20050111160331.PBKS24538.mxfep02.bredband.com@wigwam>;
          Tue, 11 Jan 2005 17:03:31 +0100
Received: from localhost ([127.0.0.1] helo=wigwam)
	by wigwam with esmtp (Exim 4.34)
	id 1CoOUu-0003D7-3f; Tue, 11 Jan 2005 17:03:56 +0100
To: hartmut.schirmer@arcormail.de
Cc: gcc@gcc.gnu.org
Subject: Re: Aw: Re: enumeration value ... not handled in switch
References: <87k6qkr0nt.fsf@wigwam.deepwood.net>
	<1F5C7ECB324CD411B58600D0B723F181779402@oehhex01.becker.de>
	<17455984.1105449856020.JavaMail.ngmail@webmail-05.arcor-online.net>
From: Daniel Brockman <daniel@brockman.se>
Date: Tue, 11 Jan 2005 17:00:00 -0000
In-Reply-To: <17455984.1105449856020.JavaMail.ngmail@webmail-05.arcor-online.net> (hartmut
 schirmer's message of "Tue, 11 Jan 2005 14:24:16 +0100 (CET)")
Message-ID: <876524q7f8.fsf@wigwam.deepwood.net>
User-Agent: Gnus/5.11 (Gnus v5.11) Emacs/21.3.50 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-SW-Source: 2005-01/txt/msg00632.txt.bz2

hartmut.schirmer@arcormail.de writes:

> Hi,
>
> [...] 
>> As it seems, this problem cannot be solved without adding an attribute
>> that declares a default case to be exceptional.
>> 
>> Consider the following code:
>> 
>>    struct fruit
>>    {
>>      enum { APPLE, MELON, ORANGE } type;
>>      ...
>>    } fruit;
> [...]
>
> Adding another fruit BANANA
>
>> [...] should fail as
>> quickly and noisily as possible:
>> 
>>    switch (fruit.type)
>>      {
>>        case APPLE:
>>          ...; break;
>>        case MELON:
>>          ...; break;
>>        case ORANGE:
>>          ...; break;
>>        default:
>>          abort ();
>>      }
>
> This may produce a warning like
> enum value 'BANANA' triggers noreturn function 'abort' in switch

I don't see how triggering noreturn functions in a switch in itself
warrants a warning.  I have lots of code that looks like this:

   void
   eat_small_fruit (struct fruit *fruit)
   {
     switch (fruit.type)
       {
         case APPLE:
           ...; break;
         case MELON:
           abort ();
         case ORANGE:
           ...; break;
       }
   }

Here, passing a melon to `eat_small_fruit' is a programmer error,
because a melon is not considered to be a small fruit.

> Creating something like __builtin_warning()
>
>     switch (fruit.type)
>       {
>         case APPLE:
>           ...; break;
>         case MELON:
>           ...; break;
>         case ORANGE:
>           ...; break;
>         default:
>           __builtin_warning();
>       }
>
> could trigger a similar warning without changing the control flow.

That suffers the same problem.  Should this trigger a warning?

   void
   eat_small_fruit (struct fruit *fruit)
   {
     switch (fruit.type)
       {
         case APPLE:
           ...; break;
         case MELON:
           __builtin_warning ();
         case ORANGE:
           ...; break;
       }
   }


Regards,

-- 
Daniel Brockman  <daniel@brockman.se>