static test failure on MAC OS X with gcc-4.9

[Juergen Reuter]

THe static test fails linking with the following error message:

libtool: link: gfortran -g -O2 -o whizard.static .libs/libmanager.o  -L/Users/reuter/local/packages/whizard-2.1.1/src/whizard-core -L/Users/reuter/local/packages/whizard-2.1.1/src -L/Users/reuter/local/packages/whizard-2.1.1/src/hepmc -L/usr/local/lib -L/usr/local/lib/gcc/x86_64-apple-darwin11.4.2/4.9.0 -L/usr/local/lib/gcc/x86_64-apple-darwin11.4.2/4.9.0/../../.. /usr/local/lib/libstdc++.a -L/usr/local/packages/gcc-4.9.0_trunk/build/x86_64-apple-darwin11.4.2/libstdc++-v3/src -L/usr/local/packages/gcc-4.9.0_trunk/build/x86_64-apple-darwin11.4.2/libstdc++-v3/src/.libs ./.libs/libstatic.a -L/Users/reuter/local/packages/whizard-2.1.1/src/models /Users/reuter/local/packages/whizard-2.1.1/src/whizard-core/.libs/libwhizard_main.a /Users/reuter/local/packages/whizard-2.1.1/src/.libs/libomega.a /Users/reuter/local/packages/whizard-2.1.1/src/omega/src/.libs/libomega_core.a /Users/reuter/local/packages/whizard-2.1.1/src/.libs/libwhizard.a /Users/reuter/local/packages/whizard-2.1.1/src/vamp/src/.libs/libvamp.a /Users/reuter/local/packages/whizard-2.1.1/src/circe1/src/.libs/libcirce1.a /Users/reuter/local/packages/whizard-2.1.1/src/circe2/src/.libs/libcirce2.a /usr/local/lib/libLHAPDF.a /usr/local/lib/libgfortran.a /usr/local/lib/libquadmath.a -lm -lstdhep -lFmcfio /usr/local//lib/libHepMC.a 
Undefined symbols for architecture x86_64:
  "std::error_category::error_category()", referenced from:
      std::future_category()  in libstdc++.a(future.o)
ld: symbol(s) not found for architecture x86_64
collect2: error: ld returned 1 exit status

seems to be only on MAC OS X, (re)testing for Linux at the moment.

[Juergen Reuter]

Ok, this function is C++11 status, cf. ​http://en.cppreference.com/w/cpp/thread/future_category .

[Juergen Reuter]

I'm going to check now whether this happens also 1. under Linux, and 2. for 4.7.3 and 4.8.0 under MAC. Just to gain some information. Could still be a problem of gcc. future_error is part of the exception handling, but the corresponding classes are not defined in <stdexcept> but in <future>.

[Juergen Reuter]

First clue: on MAC OS X with gcc <= 4.7.3 it works. Testing 4.8.0 now.

[Juergen Reuter]

And it also works with gcc 4.8.0. Only 4.9.0 (trunk) seems to fail. Hm. Somehow my suspicion is that it has to do with this here:

GCC 4.8.1 will be C++11 feature-complete [2013-04-01]
    Support for C++11 ref-qualifiers was added to the GCC 4.8 branch, making G++ the first C++ compiler to implement all the major language features of the C++11 standard. This functionality will be available in GCC 4.8.1.GCC 4.8.1 will be C++11 feature-complete [2013-04-01]
}}

[Juergen Reuter]

It seems more and more to be a problem of gcc 4.8.1 and 4.9.0 and the build of their static runtime library. I submitted a bug report, but it is only a vague one: ​http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57421

[Juergen Reuter]

I managed to check that on Linux this problem does not appear, because only for MAC OS X due to the single pass linker properties we have to explicitly link the static libstdc++.a library.

[Juergen Reuter]

It turns out that 4.8.1 is ok. So it is _not_ the C++11 standard implementation per se, but something inside the 4.9 trunk. Could be that there the C++11 standard in the static library is broken. Or it was the fault of a subversion update and a (partial) built that led to the fault. Investigating.

[Juergen Reuter]

After checking out revision r199585 and recompilation of the compiler this test works. So either it was a bogus bug or it went away. Closing for now.