Beman Dawes wrote: > Intel's 7.1 upgrade doesn't bump their release macro! It is still > reporting 700. > > I tried to update to use Intel 7.1 with VC++ 7.1, but am getting > strange results. The IDE integration doesn't seem to provide a way to > switch to the Intel compiler. Boost tests are getting weird link > failures. No std::endl for example. Either the Intel release isn't > ready for VC++ 7.1 (in spite of claims to the contrary) or I've done > something stupid. > > Which VC++ version are you integrating Intel with? Have you gotten it > to work with VC++ 7.1? >
Currently ours is configured to integrate (and be compatible with) MSVC 6.0. We didn't try anything else (yet). > Also, your setup is confusing the bjam automatic version deduction. If you mean all those 'wchar_t'-related failures, it doesn't have anything to do with the bjam version deduction - it's "boost/config/compiler/intel.hpp" that is at fault here. As per http://aspn.activestate.com/ASPN/Mail/Message/boost/1614864, a standalone Intel C++ 7.x does not automatically define _WCHAR_T_DEFINED macro, but then checking just for that one to determine whether we need to set BOOST_NO_INTRINSIC_WCHAR_T is not enough, because the former one is also getting defined in Dinkumware's <cstddef> header, so currently the success of the compilation on Intel 7.x in VC6 compatibility mode very much depends on the order of includes of various boost headers. For instance, this one-liner compiles: #include "boost/type_trais/is_integral.hpp" but this one already does not: #include <cstddef> #include "boost/type_trais/is_integral.hpp" Since the toolset automatically, through the command line, defines _NATIVE_WCHAR_T_DEFINED whenever '/Zc:wchar_t' is specified, I would say that #if BOOST_INTEL_CXX_VERSION < 700 # define BOOST_NO_INTRINSIC_WCHAR_T #else # ifndef _WCHAR_T_DEFINED # define BOOST_NO_INTRINSIC_WCHAR_T # endif #endif lines in "boost/config/compiler/intel.hpp" should look more like #if BOOST_INTEL_CXX_VERSION < 700 # define BOOST_NO_INTRINSIC_WCHAR_T #else # if !defined(_NATIVE_WCHAR_T_DEFINED) # define BOOST_NO_INTRINSIC_WCHAR_T # endif #endif but then the file's history shows that we went through quite a few iterations on this one, so I am not that sure. Aleksey _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost