Re: Flag -Wall in release?

Michèle Garoche <[email protected]> Mon, 27 Jun 2005 22:15:54 +0200
Newsgroups gmane.editors.conglomerate.devel
Message-ID <[email protected]>
Le 27 juin 2005 =E0 17:36, David Malcolm a =E9crit :

> On Sat, 2005-06-25 at 21:02 +0200, Mich=E8le Garoche wrote:
>
>> Hi,
>>
>> Would not the -Wall flag to GCC be removed in release?
>>
>> In configure.in, there is:
>>
>> BASE_CFLAGS=3D"$BASE_CFLAGS -Wall"
>>
>> It makes the console flooded with messages.
>>
>> I plan to remove it in the fink package, do you think it can do any
>> harm?
>>
>
> "-Wall" means "use all warnings" to the compiler.   I've tried hard to
> make Conglomerate compile without warnings.  The precise warnings
> generated will vary from compiler to compiler, and from environment to
> environment.  It should be harmless to remove that option.
I just thought about a flag to disable it if needed, not to remove it =20=

completely, it was wrong wording of my part.

Normaly I compile with the default -Wall and with -Werror, so that I =20
have all warnings; but for the fink package, I don't add -Werror.

The change here is that gcc-4.0 is already very, very verbose about =20
warnings, so that -Wall is about the same as -Werror, and I fear that =20=

the logs in console become so huge that it may cause weird problems.

>
> However, sometimes this is useful - occasionally I've found bugs this
> way, when an alternate compiler complains about buggy code that my
> normal compiler was accepting happily.  So I'd prefer to try to fix =20=

> all
> of the warnings, assuming this is doable.
Sure.

>
> So it would be helpful if you could capture the log of a full =20
> compile in
> your environment with -Wall (possibly by using the "script" =20
> command, if
> that's available on OS X), open a fresh bug to track this, and then
> attach the log to the bug (possibly compressed using gzip).
Ok, I'll do it with -Werror activated.

Cheers,
Mich=E8le
<http://micmacfr.homeunix.org>