This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: Binutils and GDB
On Thu, 2003-08-07 at 16:52, Daniel Jacobowitz wrote:
> On Thu, Aug 07, 2003 at 01:54:39PM +0300, Stephen Biggs wrote:
> > On Wed, 2003-08-06 at 15:53, Daniel Jacobowitz wrote:
> > > On Wed, Aug 06, 2003 at 03:05:27PM +0300, Stephen Biggs wrote:
> > > > Greetings all,
> > > >
> > > > I apologize for what will probably seem a hopelessly clueless and newbie
> > > > question, but I am stuck, so here goes:
> > > >
> > > > I notice that the GDB source tree has a lot of what seems to be almost
> > > > identical code in common with the binutils source tree. I have made
> > > > some changes to the binutils 2.14 source tree, specifically in the BFD
> > > > and opcodes directories that I wish to integrate into GDB. How do I do
> > > > this with the minimum amount of effort? Is there a way to tell the GDB
> > > > configure to not configure the GDB's bfd, rather use another already
> > > > built BFD library? How, if so?
> > >
> > > No, GDB can't use the system BFD. I recommend just applying the patch.
> > > The directory is common to both projects, but gdb and binutils branch
> > > at different times.
> > >
> > But, this is a big mess, no? That means that any changes in the system
> > binutils BFD have to be reflected in the GDB BFD and back-patched, which
> > they seem NOT to be... how does this work at all?
>
> Eh?
>
> The master sources for binutils and GDB live in the same CVS
> repository. So the masters are always in sync. Distributors have to
> patch both copies if they need local patches - but in general, they
> don't.
An example off the top of my head is the change in the latest version
(or a couple of versions before, I don't know exactly) of the BFD where
all references to "boolean" were changed to "bfd_boolean" and
"true/false" to "TRUE/FALSE". This did NOT make it into the GDB version
and it is a big change for portability, isn't it? I don't understand
how you can say that the masters are always in sync?