This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: change behavior of elfos.h slightly
On Fri, May 05, 2000 at 12:18:45PM +0200, Franz Sirl wrote:
> >We're trying to enforce it for new ports and when major reconstruction is
> >done for existing ports. But we haven't forced every port to switch when
> >dealing with minor problems.
>
> Hmm, I thought the NetBSD stuff was a major reconstruction like OpenBSD was
> sometime ago, and AFAI remember the OpenBSD maintainer was tweaked quite a
> bit before his patches were accepted :-).
>
> My experience with the newppc-branch let's me think that the extra work is
> really worth the effort. It increases maintainability and readability a lot
> and may even uncover latent bugs.
So, for what it's worth, I appreciate the goal, here. I agree wholeheartedly
that the GCC configuration headers need to be cleaned up and clarified :-)
However, here are my issues:
(1) Right now, I'm primarily concerned with getting basic NetBSD
support in the GNU tools up-to-par with other platforms.
(2) As a secondary concern, I would like to make the NetBSD
support in the GNU tools as modular as possible, so that
common definitions aren't needlessly duplicated in multiple
files. This is a maintenance nightmare considering that
NetBSD supports 10 CPU architectures and 3 different object
formats (a.out on some, ELF on some, ECOFF on some, some
support multiple formats, e.g. a.out+ELF or ECOFF+ELF, tho
the intent is to move everyone to ELF).
(2) Many of the header files I have to work with on some arch's
(like, e.g. the i386) aren't up-to-snuff with how you want
the new scheme to work, and so I'm sort of backed into a
corner.
In any case, I'm certainly not trying to make things any *worse*. I'll
do my best to work within your contraints, but I would appreciate a little
flexibility.
--
-- Jason R. Thorpe <thorpej@zembu.com>