This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: GCC-3.0.4 (pre) on AIX requires pthreads
- From: Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>
- To: dje at watson dot ibm dot com
- Cc: libstdc++ at gcc dot gnu dot org
- Date: Fri, 22 Feb 2002 17:48:03 -0600 (CST)
- Subject: Re: GCC-3.0.4 (pre) on AIX requires pthreads
- References: <200202222324.SAA29228@makai.watson.ibm.com>
- Reply-to: rittle at labs dot mot dot com
> [...] I am not disagreeing with your original patch, just pointing out
> that there may be a simpler alternative.
Simpler is better. The other idea can be saved and pulled up if we
can undertake additional global rearchitecture. Even I dreaded
testing my approach on enough targets to satisfy everyone.
Loren> David, would you agree that this is a safe way to learn the required
Loren> dependent model files that must be installed on AIX:
Loren> ; sed -n 's,^#include.*"\(gthr-.*\)"$,\1,p' gthr-aix.h
Loren> gthr-posix.h
Loren> gthr-single.h
> Yes, this seems to be sufficient for the cases at hand.
OK.
> Am I correctly assuming that gthr-aix.h will be recognized as
> glibcpp_thread_h and installed as gthr-default.h? That should avoid a
> requirement to teach gthr.h about AIX threads or having GCC emit
> -D_PTHREADS to trigger gthr.h machinery.
gthr-aix.h will be so recognized if aix is what is reported as the
thread model. Agreed on the second point. (Note to others reading
the e-mail: The per-system macros are OK if not seen by the user. One
big problem with the thread stuff before gcc 3.0 release was the ad
hoc nature of the command-line flags across ports.)
Regards,
Loren