"Arkadiy Vertleyb" <vertleyb@hotmail.com> wrote in message
"David Abrahams" <dave@boost-consulting.com> wrote
Unfortunately there are still problems with vc-8:
..\..\..\..\boost/sequence/detail/typeof_add.hpp(15) : error C2893: Failed to specialize function template 'char (&boost::type_of::size(const T
&))[boost::mpl::size<encode_type<boost::type_of::vector0<>,T>::type>::value]
'
With the following template arguments: 'T' ..\..\..\..\boost/sequence/detail/typeof_add.hpp(16) : see
reference to class template instantiation 'boost::sequence::detail::typeof_add<T,U>' being compiled
See the updated archive.
I've run into this problem before. Looks like it's present only in vc8 beta. Seems to be "fixed" if make<T>() is replaced with T() -- of course then we have problems with non-default-constructible classes. *(T*)0 doesn't work either.
Doesn't look like a Typeof problem, although Typeof may be responsible for driving the compiler crazy, after which it chokes.
Seems to be a question to Microsoft...
After looking at this more closely, I believe the problem can't always be "cured" in the way I suggested. Also it can be reproduced on both vc71 and vc8 beta. Here is the minimal example: // ---- #include "boost/mpl/vector.hpp" #include "boost/mpl/size_t.hpp" template<class T> T make(); template<class T> struct et { typedef boost::mpl::vector1<int> type; }; template<class T> char(&sz(const T&))[ boost::mpl::size< typename et<T>::type >::value ]; template<class T, class U> struct add { typedef boost::mpl::vector1< boost::mpl::size_t<sizeof(sz(make<T>() + make<U>()))> //boost::mpl::size_t<sizeof(sz(T() + U()))> -- doesn't help here > type; }; int main() { return 0; } // ---- This example compiles fine with GCC, but both MS compilers choke on it. OTOH, our ODR test uses BOOST_TYPEOF_TPL(T() + U()) in a few different contexts, and it works fine on both MS compilers. Do Microsoft people read this list, or should I post it elsewhere? Regards, Arkadiy