here’s a template class with default template parameter:

#include <cstdio>
constexpr int d = 1;

template <int a = d> struct test;

template <> struct test<1> {
  test<1>(int a) { printf("%dn", a); }
};
template <> struct test<2> {
  test() { printf("%dn", 2); }
};
int main() { test t{1}; }

oddly enough, if d=1 it won’t compile, while d=2 it’s fine. if I want use it when d=1, I have to do it like this: test<> t{1};. isn’t it kind of counter-intuitive?