This is the mail archive of the
libstdc++@sources.redhat.com
mailing list for the libstdc++ project.
Re: AIX atomicity.h support
- To: David Edelsohn <dje at watson dot ibm dot com>
- Subject: Re: AIX atomicity.h support
- From: Phil Edwards <pedwards at disaster dot jaj dot com>
- Date: Thu, 28 Sep 2000 15:58:06 -0400
- Cc: libstdc++ at sources dot redhat dot com, Mark Mitchell <mark at codesourcery dot com>, Geoff Keating <geoffk at cygnus dot com>
- References: <bkoz@redhat.com> <200009281453.KAA08956@mal-ach.watson.ibm.com>
On Thu, Sep 28, 2000 at 10:53:28AM -0400, David Edelsohn wrote:
>
> I also needed to define the following macros and typedefs:
>
> #define __P(p) p
> #define __PMT(p) p
> #define __THROW /**/
> typedef long long __off_t;
> typedef long long __off64_t;
> typedef ssize_t __ssize_t;
>
> AIX has /usr/include/sys/cdefs.h but it does not define __P and
> __PMT, so _G_HAVE_CDEFS succeeds incorrectly in libio.h.
Join the glibc-less club, brother. :-)
These are almost the exact same problems that Solaris experiences.
(Although Sol8 doesn't even have a sys/cdefs.h anywhere. And it defines
its own __P without checking for previous #definitions.)
I posted some truly ugly hacks to the list a few weeks back. If you were
willing to disable wide character support, you could get a pseudo-working
Solaris build, useful for testing other things. The patches were based
on bringing a sys/cdefs.h from a Linux box into the library and making
slight changes, which is probably not a viable solution.
I strongly suspect that solving these problems for either OS will fix most
of the problems for the other OS.
Phil
--
pedwards at disaster dot jaj dot com | pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools. Fools are protected by more capable fools.