Warnings about rcs_id strings: let's settle this

[Joe Buck]

On Fri, May 02, 2003 at 07:19:31PM +0200, Gabriel Dos Reis wrote:
> Phil Edwards <phil@jaj.com> writes:
> 
> | For people who haven't been reading that other exploding thread:  here
> | are some actual data from actual users:
> 
> Whatever is decided for the unused variable, we need to say something
> clear about -Wall.  This is not the first time we're getting
> discusions about -Wall because a new warning was added (or removed).

We should encourage developers to use -Wall, and we should expect that
"good" code will be clean or mostly clean with -Wall.  When a new warning
is added, it should go into -Wall if we're pretty sure that the things it
detects are likely to be serious problems.  We cannot promise that code
that is clean with -Wall will always remain so, but when we know that we've
done something that will make -Wall much noisier, we should consider a
rethink.

So that we "eat our own dogfood", gcc itself should be -Wall-clean.  If
we find that we can't achieve this, it indicates problems with -Wall.