-Wswitch and default case

-Wswitch and default case

[Tom Tromey]

I've run into a case where I'd like to use -Wswitch but also have a
default case in my switch.

Suppose I have a switch for something like a bytecode interpreter (my
actual situation is the Java bytecode verifier).  I have a large
number of bytecodes, which are in an enum.  I'd like to make sure I
don't forget one of the opcodes in one of my switches, so I'd like to
use -Wswitch.  But at the same time if there is a bogus bytecode, I'd
like to catch that in a `default' case.  Unfortunately the presence of
a default case eliminates whatever utility -Wswitch might have
provided.

I know somebody has wanted this before:

    http://gcc.gnu.org/ml/gcc/2000-11/msg00854.html

Has anybody tried to implement this change?
If not I'll submit a feature request.

Tom

Re: -Wswitch and default case

[Hartmut Schirmer]

On Fri, 23 Nov 2001, Tom Tromey wrote:
>I've run into a case where I'd like to use -Wswitch but also have a
>default case in my switch.
>
[...]
>Has anybody tried to implement this change?
>If not I'll submit a feature request.

I like this feature. It´s hard to write robust code
in C using enum´s and switches and gcc diagnostic could
be much better here.

One could introduce a command line option like -Wswitch-all
or/and enable this option in the source for every default that
should be hidden by adding an attribute to default or use a
special version like __default / __silent_default /  ...

Hartmut

Re: -Wswitch and default case

[Alexandre Oliva]

On Nov 23, 2001, Tom Tromey <tromey@redhat.com> wrote:

> Has anybody tried to implement this change?

Not as far as I know.

> If not I'll submit a feature request.

I suggest making the default: label with attribute((unused)) would be
a reasonable way to tell the compiler that (i) the default is not
supposed to be used very often (which helps us get better code from
basic-block reordering and branch probability analysis) and (ii) it's
there just to catch erroneous cases, so the compiler should still warn
if any of the enum members doesn't have a corresponding case.  Hmm,
but perhaps this is overloading attribute((unused)) too much?

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

-Wswitch and default case

[Claus Fischer]

It would be nice to have such a warning (reporting missing
cases in the presence of the default: label) but it should
be distinct from -Wswitch, IMHO. Give it a separate name,
if you do it.

Claus

-- 
Claus Fischer <claus.fischer@clausfischer.com>
http://www.clausfischer.com/