On Feb 2, 2013, at 9:17 AM, Christian Henning <chhenning@gmail.com> wrote:

Hi there,

Can someone shed some light on the following compiler errors:

c:\boost\boost\gil/channel_algorithm.hpp:54:85: error: non-type template argument evaluates to -1, which cannot be narrowed to type 'unsigned long long' [-Wc++11-narrowing] struct unsigned_integral_max_value : public mpl::integral_c<UnsignedIntegralChannel,-1> {};

^ c:\boost\boost\gil/channel_algorithm.hpp:204:19: note: in instantiation of template class 'boost::gil::detail::unsigned_integral_max_value<unsigned long long>' requested here if (src > unsigned_integral_max_value<uintmax_t>::value - div2)

I only get those in c++11 mode. The code to trigger them is:

The error message seems pretty straightforward. The code is trying to create value of type "unsigned long long" with the value "-1". This is (obviously) not a problem with your code, but rather a problem in Boost.Gil, related to the increased checking in Clang 3.2. Clang is nice enough to tell you exactly how to disable this warning: add "-Wnoc++11-narrowing" to your CXXFLAGS (or however your environment lets you pass arguments to the compiler). Alternately, you could change c:\boost\boost\gil/channel_algorithm.hpp, line 54 from: struct unsigned_integral_max_value : public mpl::integral_c<UnsignedIntegralChannel,-1> {}; to: struct unsigned_integral_max_value : public mpl::integral_c<UnsignedIntegralChannel,UnsignedIntegralChannel(-1)> {}; or, even better, to: struct unsigned_integral_max_value : public mpl::integral_c<UnsignedIntegralChannel,std::numeric_limits<UnsignedIntegralChannel>::max()> {}; Either of those should solve your problem, I think (depending on your standard library) -- Marshall P.S. Please open a ticket at svn.boost.org so we can be sure to fix this in a future release.