[Patch] [add changelog] reduce template instantiation depth in <variant>

[Barrett Adair]

On Sun, Nov 13, 2016 at 3:08 AM, Tim Song <t.canens.cpp@gmail.com> wrote:
> On Sat, Nov 12, 2016 at 3:48 PM, Barrett Adair
> <barrettellisadair@gmail.com> wrote:
>> On Sat, Nov 12, 2016 at 2:39 PM, Tim Shen <timshen@google.com> wrote:
>>> On Sat, Nov 12, 2016 at 4:28 AM, Daniel Krügler
>>> <daniel.kruegler@gmail.com> wrote:
>>>> 2016-11-12 10:04 GMT+01:00 Barrett Adair <barrettellisadair@gmail.com>:
>>
>> Since there are no side-effects, and the metafunctions in play cannot
>> SFINAE, I don't believe the semantics are actually affected by
>> short-circuiting in this case.
>
> The issue is that the fold expression requires actually instantiating all the
> is_meow_constructible &c. specializations (which in turn may trigger extra
> instantiations) while __and_ only instantiates those needed for the result.
> The extra instantiations can be expensive in terms of compile time but this,
> of course, depends on the types involved. Some benchmarks would be nice.

Here is a benchmark that I made with Metabench: http://jsbin.com/qalegikowi

It appears that the effect on compile-time performance is negligible at worst.
My original interest in contributing to <variant> was to reduce the compile-time
complexity by number of types. It would be nice if to have a linear
implementation, but I don't know yet whether that is possible.

Here is the Ruby template I used for the benchmark:

#include <variant>

int main(int argc, char**) {

#if defined METABENCH

    struct T { int i; };

    <% (0..n).each do |i| %>
    struct T<%= i %> { int i; };
    <% end %>

    using var = std::variant< T, T<%= (0..n).to_a.join(', T') %> >;

    var x;
    x = T{argc};
    std::visit([](auto v){ return v.i; }, x);

#endif
}