Two builds of root on Mac OS with gcc and clang

Hi all. For starters: I’m running Mac OS 10.8.5.

I’m trying to build and install two separate copies of root on my machine – one built with gcc/g++ (I’m trying for version 6.04.08, with gcc 5.1.0), and one built with the native clang (I’ve got version 5.34.34 working, and I’d really like to keep this one as version 5. For clang, I have: Apple LLVM version 5.1 (clang-503.0.40) (based on LLVM 3.4svn), and this seems to work without any issues.). I have two (possibly related) problems with this:

Firstly: when I recently re-built root 5 with clang, it broke my previously functional (gcc-based? I think?) installation of root 6. So I suspect that I’m not keeping the various files as separate as I might like to think I am.

Secondly: I can’t actually finish installing my gcc build of root 6. I get all the way through “make” without any issues, but when I then have it “make install”, it comes back with a linker error:

ld: symbol(s) not found for architecture x86_64 collect2: error: ld returned 1 exit status make: *** [bin/rootcling] Error 1 (I can post the full output if anyone thinks it would be useful.)

So … am I doing something obviously wrong here? Is there a better way to tell it to link to gcc libraries and not clang libraries? Is it looking through old clang-based root files stored somewhere non-obvious?

So here’s how I ./configure-ed root 6:

./configure --fail-on-missing --prefix=/Users/spiffyzha/Packages/root6 --enable-soversion --enable-rpath --enable-x11 --enable-xft --disable-gsl-shared --enable-roofit --with-cc=gcc-5.1.0 --with-cxx=g++-5.1.0 --with-gcc-toolchain=/usr/gcc-5.1.0/bin --disable-cocoa --enable-builtin-freetype --gminimal --disable-fink --with-ld=g++-5.1.0
and I get back:

Enabled support for builtin_freetype, builtin_llvm, explicitlink, roofit, rpath, search_usrlocal, shared, soversion, x11, xft.

…And here’s what I did for root 5:

And the result was:

Hi,

Do you still have this problem? If so, [quote](I can post the full output if anyone thinks it would be useful.) [/quote]yes, we would need to see the full error output.

Thanks,
Philippe.