http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58713
--- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> --- (In reply to Daniel Krügler from comment #3) > guess is that this would also improve the diagnostics in this case, is that > right? Even if it worked (which it doesn't, as Marc also discovered) it would only help a little bit, because G++ still prints all the candidates even if they are not viable due to substitution failure. (Usually I love this feature, but when there are a very large number of overloads, such as anything using operator<<, the diagnostics can be overwhelming.) (In reply to Daniel Krügler from comment #6) > The diagnostics of my gcc current 4.9.0 is > > <quote> > main.cpp||In function 'int main()':| > main.cpp|50|error: no match for 'operator<<' (operand types are > 'xstd::OStream<>' and 'A')| > main.cpp|50|note: candidates are:| > main.cpp|9|note: xstd::OStream<CharT>& xstd::OStream<CharT>::operator<<(int) > [with CharT = char]| > main.cpp|9|note: no known conversion for argument 1 from 'A' to 'int'| > main.cpp|10|note: xstd::OStream<CharT>& > xstd::OStream<CharT>::operator<<(short int) [with CharT = char]| > main.cpp|10|note: no known conversion for argument 1 from 'A' to 'short > int'| > main.cpp|11|note: xstd::OStream<CharT>& > xstd::OStream<CharT>::operator<<(double) [with CharT = char]| > main.cpp|11|note: no known conversion for argument 1 from 'A' to 'double'| > main.cpp|12|note: xstd::OStream<CharT>& > xstd::OStream<CharT>::operator<<(const void*) [with CharT = char]| > main.cpp|12|note: no known conversion for argument 1 from 'A' to 'const > void*'| > main.cpp|36|note: template<class CharT, class T, typename > std::enable_if<xstd::is_ostreamable<CharT, const T&>::value, bool>::type > <anonymous> > xstd::OStream<CharT>& xstd::operator<<(xstd::OStream<CharT>&&, > const T&)| > main.cpp|36|note: template argument deduction/substitution failed:| > main.cpp|35|error: no type named 'type' in 'struct std::enable_if<false, > bool>'| > main.cpp|35|note: invalid template non-type parameter| > </quote> This does look better. I thought I'd experimented with a similar trait(*) and found the results disappointing. (*) https://gitorious.org/redistd/redistd/source/33d8d3a4f49da25e71ca87aab674d3b88500d302:include/redi/printers.h
