Re: experimental gcc-2.95.3 and gcc-2.97 (20001224)
Daniel Jacobowitz writes:
> On Mon, Dec 25, 2000 at 10:44:09AM +0100, Matthias Klose wrote:
> > Daniel Jacobowitz writes:
> > > On Mon, Dec 25, 2000 at 06:18:33AM +0100, Matthias Klose wrote:
> > > > Currently the gcc-2.95.2 package has binaries {c89,gcc,gcov}{,-2.95}.
> > > > g++-2.95.2 has binaries g++-2.95 and g++. gcc-2.97 has the gcc-2.97
> > > > binary and g++-2.97 the g++-2.97 binary. Both gcc-2.9x and g++-2.9x
> > > > packages provide an alternative cc/c++. By using gcc/g++ you get the
> > > > default compiler per architecture.
> > >
> > > So cc can be the one or the other, but gcc will be fixed by
> > > architecture? That doesn't seem to make sense. A lot of packages just
> > > use cc to build.
> >
> > Agreed. Then only the default-gcc-for-arch package provides the
> > alternative cc/c++. Or can we drop the handling of cc/c++ by
> > alternatives? For f77 that's not a solution, because f2c is the
> > preferred f77 on some platforms. pc is provided by free-pascal as
> > well. java isn't yet in the game.
>
> I'm all for having the default-gcc-for-arch package do it.
>
> Are we calling that package "gcc"? I would hope so - keep dependencies
> simple.
Source package gcc-defaults, binary-arch packages cpp, gcc, g++ and maybe
g77, gcj and gpc. For gcc: gcc is a symlink to gcc-x.yz and the
package provides the alternative cc (pointing to gcc). Will this work?
What about packages like colorgcc?
Reply to: