On Dec 28, 2008, at 1:25 AM, David Abrahams wrote:
The conundrum here is that if we use that technique for tuple, then a tuple of empty types can never be empty itself. I'm starting to get ill just thinking about a generic framework for composing "logically empty" types that undoes this "EBO erasure" effect when necessary.
Perhaps I'm misunderstanding. Reverting from English to C++ in the hopes of clarification. :-) The C++0X program and its output below look quite reasonable to me: #include <tuple> #include <type_traits> #include <cstdio> struct empty1 {}; struct empty2 {}; struct empty3 {}; int main() { typedef std::tuple<empty1> T1; std::printf("is_empty<tuple<empty1>> = %d\n", std::is_empty<T1>::value); typedef std::tuple<empty1, empty2> T2; std::printf("is_empty<tuple<empty1, empty2>> = %d\n", std::is_empty<T2>::value); typedef std::tuple<T2, empty3> T3; std::printf("is_empty<tuple<tuple<empty1, empty2>, empty3>> = %d \n", std::is_empty<T3>::value); typedef std::tuple<int, T3> T4; std::printf("is_empty<tuple<int, tuple<tuple<empty1, empty2>, empty3>>> = %d\n", std::is_empty<T4>::value); std::printf("sizeof(int) = %zu, sizeof(tuple<int, tuple<tuple<empty1, empty2>, empty3>>) = %zu\n", sizeof(int), sizeof(T4)); } is_empty<tuple<empty1>> = 1 is_empty<tuple<empty1, empty2>> = 1 is_empty<tuple<tuple<empty1, empty2>, empty3>> = 1 is_empty<tuple<int, tuple<tuple<empty1, empty2>, empty3>>> = 0 sizeof(int) = 4, sizeof(tuple<int, tuple<tuple<empty1, empty2>, empty3>>) = 4 I currently favor the "shallow hierarchy" design Doug outlined in c+ +std-lib-18989. Though EMO isn't shown in that description, it is pretty simple to add to that design. -Howard