This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author trent
Recipients amaury.forgeotdarc, belopolsky, jcea, meador.inge, skrah, theller, trent
Date 2012-10-18.14:58:23
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <20121018145816.GA92513@snakebite.org>
In-reply-to <1350571660.3.0.00715010586627.issue16275@psf.upfronthosting.co.za>
Content
On Thu, Oct 18, 2012 at 07:47:40AM -0700, Stefan Krah wrote:
> 
> Stefan Krah added the comment:
> 
> --with-system-ffi should probably solve some problems, see #12927.

    This isn't an option on Solaris 10, as it doesn't ship with a
    "system ffi", unfortunately.  (That being said, after building
    the latest libffi with Sun C, I do use --with-system-ffi in
    order to override the building of Modules/_ctypes/libffi.)
>
> The summary of that issue is that the unpatched libffi is not
> compatible with suncc and any issues should be reported upstream
> (at least that was Meador's and my own opinion).

    Sun CC is supported by libffi for SPARC only.  You're right
    regarding x86/x64 though; libffi explicitly doesn't support
    using anything other than gcc on Solaris if it's x86.

    (Unfortunately, my only SPARC boxes are running Solaris 9
     and 10.  I don't have new-enough SPARC CPUs that can run
     Solaris 11.  If I did, this would probably be moot, as I
     could just use the system libffi Sun provide.)

    Side note: on my Solaris 11 x64 box, --with-system-ffi works
    (once -I and -L are tweaked), but test_ctypes still fails if
    you're building with Sun CC.
History
Date User Action Args
2012-10-18 14:58:24trentsetrecipients: + trent, theller, jcea, amaury.forgeotdarc, belopolsky, skrah, meador.inge
2012-10-18 14:58:23trentlinkissue16275 messages
2012-10-18 14:58:23trentcreate