Variadic template wrapper around tuple

Asked Apr 03 '19 at 08:26

Active Apr 03 '19 at 08:31

Viewed 109 times

I am currently in the process of updating my language knowledge (to C++17). I was trying to get this code to compile using gcc and couldn't figure it out so I tried MSVC and clang and both seem okay with this code -- is there anything wrong with this bit of code?

gcc seems to be complaining about instantiating a tuple<> with no template parameters which is odd:

prog.cc: In instantiation of 'C::C(U&& ...) [with U = {const A&, const B&}; T = {}]': prog.cc:32:14: required from here prog.cc:12:49: error: no matching function for call to 'std::tuple<>::tuple(const A&, const B&)' 12 | : container( std::forward(params)... ) | ^

#include <tuple>

struct A {};
struct B {};

template<typename... T>
struct C
{
    template<typename... U>
    C(U&&... params)
        : container( std::forward<U>(params)... )
    {}

    std::tuple<T...> container;
};

template <class... T>
C(T&&...)->C<std::decay_t<T>...>;

int main(void)
{
    const A a;
    const B b;
    C c2(a, b);  // <-- this fails on gcc, okay on msvc and clang
}

edited Apr 03 '19 at 08:31

lubgr

37,368
3
66
117

asked Apr 03 '19 at 08:26

squashed

@xaxxon the issue is that he wants to know why it won't compile on gcc :) – Richard Hodges Apr 03 '19 at 08:34
Which version of gcc? – francesco Apr 03 '19 at 08:36
@francesco fails with 8.3 https://godbolt.org/z/uXQA_y – xaxxon Apr 03 '19 at 08:39
the problem seems to be: `: In instantiation of 'C::C(U&& ...) [with U = {const A&, const B&}; T = {}]':` <== T is empty, so it's an empty tuple type. Is there some sort of conflict of deduction guides or something? I don't know – xaxxon Apr 03 '19 at 08:41
1

If you add a SFINAE friendy error, it seems to work (`template> C(U&&... params)`, or even `std::enable_if_t`) – Artyer Apr 03 '19 at 08:44
Removing deduction guide make clang fails a similar way...[Demo](http://coliru.stacked-crooked.com/a/b5067cd6273a06e5) – Jarod42 Apr 03 '19 at 08:44
unusually, this looks like a bug in gcc – Richard Hodges Apr 03 '19 at 08:44
@RichardHodges gcc has tons of bugs – xaxxon Apr 03 '19 at 08:45
3

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80438 – cpplearner Apr 03 '19 at 08:50
@cpplearner: But the given workaround did not work. So I believe that is an additional problem. So it seems to be helpfull to create a new bug entry or add the code example from here to the exsisiting one! – Klaus Apr 03 '19 at 08:54
@xaxxon: Not with unsiversal reference type. If you use `template < class... T> C(T&...)->C...>;` it works also. So for me it seems a different thing – Klaus Apr 03 '19 at 08:59
https://godbolt.org/z/zo6oLT – xaxxon Apr 03 '19 at 09:00
@xaxxon: Thanks! Had a type in my local example.... works! – Klaus Apr 03 '19 at 09:41
1

Unrelated, but a better way to write the deduction guide is just `template C(T...) -> C` - the template deduction rules already do the decaying for you. – Barry Apr 03 '19 at 11:41

Variadic template wrapper around tuple

0 Answers0