https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66059
Bug ID: 66059
Summary: make_integer_sequence should use a log(N)
implementation
Product: gcc
Version: 4.9.0
Status: UNCONFIRMED
Keywords: compile-time-hog
Severity: normal
Priority: P3
Component: libstdc++
Assignee: unassigned at gcc dot gnu.org
Reporter: rhalbersma at gmail dot com
Target Milestone: ---
The following program fails to compile for g++ 4.9.0 and further
#include <utility>
int main()
{
constexpr auto N = 1024;
using T = std::make_integer_sequence<int, N>;
static_assert(T::size() == N, "");
}
fatal error: template instantiation depth exceeds maximum of 900 (use
-ftemplate-depth= to increase the maximum)
Note that Clang with -stdlib=libc++ can compile integer sequences of up to one
million for machines with enough memory, and around 65K for online compilers
such as wandbox.com. Clang + libstdc++ requires -ftemplate-depth=N for N beyond
256.
Changing g++ default template instantiation depth beyond 900 is not a viable
solution. It simply masks the problem of excessive template instantiations.
Instead, the implementation of make_integer_sequence<T, N> should be made to
scale as log(N). See e.g. http://stackoverflow.com/a/17426611/819272
