This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [basic-improvements] N-way choice followup
- From: Zack Weinberg <zack at codesourcery dot com>
- To: Neil Booth <neil at daikokuya dot co dot uk>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 4 Oct 2002 12:39:02 -0700
- Subject: Re: [basic-improvements] N-way choice followup
- References: <20020930003645.GA6130@codesourcery.com> <20021004191832.GI22330@daikokuya.co.uk>
On Fri, Oct 04, 2002 at 08:18:33PM +0100, Neil Booth wrote:
> Zack Weinberg wrote:-
>
> > I would have liked to get rid of the extra parser and just set the
> > 'switch validated' flag bits as part of normal processing, but this
> > breaks the existing invariant that switches meaningful to language A
> > are silently ignored when compiling language B.
>
> I didn't preserve that with front-end specific options across C/ObjC/C++
> when I did c-opts.c. I originally made it an error, but was requested
> to downgrade it to a warning for a while at least. IMO the only reason
> things are the way they are is that being precise (particularly in the
> driver) is too hard in general. It's easy for the C family options,
> so I did it.
Well, when I say 'breaks the existing invariant', I mean 'you get
floods of warnings during a bootstrap if you try to do it the other
way'. Specifically, the libgcc build protocol applies CFLAGS to a
bunch of on-the-fly-generated .s files, and CFLAGS contains
preprocessor switches...
zw