Re: [eigen] 3.0-RC1 released

[Rhys Ulerich <rhys.ulerich@xxxxxxxxx>]

> http://bitbucket.org/eigen/eigen/get/3.0-rc1.tar.bz2
>
> Please give this a last round of testing.

I probably should have asked earlier-- is it considered okay that
unsupported/test/CMakeFiles/openglsupport.dir/openglsupport.cpp.o breaks the
'make check' target?  All of the beta releases have consistently tripped
    /workspace/rhys/tmp/eigen-eigen-b6456624eae7/unsupported/test/openglsupport.cpp:
In function ‘void test_openglsupport()’:
    /workspace/rhys/tmp/eigen-eigen-b6456624eae7/unsupported/test/openglsupport.cpp:323:8:
error: ‘GLEW_ARB_gpu_shader_fp64’ was not declared in this scope
on my system (GNU and Intel alike) and I just kept assuming it would be fixed
prior to 3.0.

I do get a
    -- Found Glew: /usr/lib/libGLEW.so
at cmake configuration time and appear to have /usr/include/GL/glew.h on
my system.

Any possibility the #ifdef GL_ARB_gpu_shader_fp64/#endif lines could be moved
outside the if(GLEW_ARB_gpu_shader_fp64) test?  I'm unfamiliar with GL and
don't know if that would be verboten, but it would solve the 'make check' break
on my system.  Something like

    #ifdef GL_ARB_gpu_shader_fp64
    if(GLEW_ARB_gpu_shader_fp64)
    {
        // ...
    }
    else
    #endif
      std::cerr << "Warning: GLEW_ARB_gpu_shader_fp64 was not tested\n";

instead of the currently problematic

    if(GLEW_ARB_gpu_shader_fp64)
    {
      #ifdef GL_ARB_gpu_shader_fp64
      // ...
      #endif
    }
    else
      std::cerr << "Warning: GLEW_ARB_gpu_shader_fp64 was not tested\n";

- Rhys