This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [RFA] fix for utils.c bool problem
On Fri, Feb 08, 2002 at 11:51:10AM -0500, Andrew Cagney wrote:
> >This is the patch I alluded to earlier today on binutils@. It's not
> >>quite complete, since TUI is also affected; I've attached mine. I'm
> >>quite surprised that this fixes the problem for you without the
> >>corresponding bfd patch that I haven't committed yet!
> >
> >
> >No one objected, so I've committed this. As far as I know GDB should
> >build everywhere it used to before I started playing with bool, as long
> >as you update both BFD and GDB. If I'm wrong, please let me know.
>
>
> Dan I couldn't see the point of the patch. The immediate problem -
> bfd.h including <stdbool.h> was fixed. The next step, I thought, was
> fix bfd.h. It isn't hurting in a major way and definitly breaks GDB's
> coding style.
Sorry if I missed an objection. I was a little out of it yesterday.
The immediate problem was -not- fixed. To start back at the beginning
again:
- On my machine, running a current Debian system, <curses.h> includes
<stdbool.h>. We have to live with that. There's nothing I can
particularly do about it.
- The way I tried to fix this was by also using stdbool.h if it was
available. But <stdbool.h> conflicts with an awful lot of existing
code. This is unfortunate, and this is where the proper fix lies,
IMO.
- The way I settled on fixing this, and committed, was to use
<stdbool.h> if something included before bfd.h had already brought
it in. This appears to work in all cases.
I strongly want to avoid leaving GDB unbuildable on this class of
systems. I don't have any particular attachment to my patch. I would
love to revert it, as soon as there is an alternative solution in
place.
--
Daniel Jacobowitz Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer