This is the mail archive of the
libffi-discuss@sourceware.org
mailing list for the libffi project.
Re: Please help test libffi 3.0.11-rc3
On Apr 6, 2012, Anthony Green <green@moxielogic.com> wrote:
>
> I've uploaded release candidate 3 for testing here:
> ftp://sourceware.org/pub/libffi/libffi-3.0.11-rc3.tar.gz
>
> Please test it on whatever systems you have at your disposal, and
> update the testing wiki here:
> http://www.moxielogic.org/wiki/index.php/Libffi_3.0.11
All tests pass on OSX Lion 10.7.3 using XCode-4.3.2 when I build
the software with clang or llvm-gcc with one major caveat. I had
to runtest with gcc-4.6.
It passes all when I build with clang-3.1.318 and test with gcc-4.6.
It passes all when I build with llvm-gcc-4.2.1 (llvm 2336.9) and
test with gcc-4.6.
It still fails if I runtest with clang or llvm-gcc. So why does that happen?
If it can build all the libraries at -Os using only those compilers,
why can't the tests pass using those compilers? Why am I forced
to install gcc-4.6? OSX doesn't come with gcc anymore, and building
that to make your testsuite work was not fun. But I did it because
I'm not going leave this as a mess on my part. OSX uses
DYLD_LIBRARY_PATH, not LD_LIBRARY_PATH, but as I mentioned
I am not a programmer. So I have no idea what to look for. In a
previous message you said my compilers have optimization problems.
If that was the case, why did the test at -O0 fail? I'm sure you'll
just ignore this anyway, but I did my part.