sizeof variadic template (sum of sizeof of all elements) - c++

Considering the following function :
template<typename... List>
inline unsigned int myFunction(const List&... list)
{
return /* SOMETHING */;
}
What is the most simple thing to put instead of /* SOMETHING */ in order to return the sum of sizeof all arguments ?
For example myFunction(int, char, double) = 4+1+8 = 13

In C++17, use a fold expression:
template<typename... List>
inline constexpr unsigned int myFunction(const List&... list)
{
return (0 + ... + sizeof(List));
}

unsigned myFunction() {return 0;}
template <typename Head, typename... Tail>
unsigned myFunction(const Head & head, const Tail &... tail) {
return sizeof head + myFunction(tail...);
}

Based off of this comment and the following comments on the question, you could use this (note: completely untested)
std::initializer_list<std::size_t> sizeList = {sizeof(List)...}; //sizeList should be std::initializer_list, according to the comments I linked to
return std::accumulate(sizeList.begin(), sizeList.end(), 0);

Two years late but an alternative solution guaranteed to be computed by the compiler (if you don't mind the different syntax):
template < typename ... Types >
struct SizeOf;
template < typename TFirst >
struct SizeOf < TFirst >
{
static constexpr auto Value = (sizeof(TFirst));
};
template < typename TFirst, typename ... TRemaining >
struct SizeOf < TFirst, TRemaining ... >
{
static constexpr auto Value = (sizeof(TFirst) + SizeOf<TRemaining...>::Value);
};
Used as constexpr std::size_t size = SizeOf<int, char, double>::Value; // 4 + 1 + 8 = 13

Here's a template way:
#include <iostream>
template<typename T, typename ...Ts>
class PPackSizeOf
{
public:
static const unsigned int size = sizeof(T) + PPackSizeOf<Ts...>::size;
};
template<typename T>
class PPackSizeOf<T>
{
public:
static const unsigned int size = sizeof(T);
};
template<typename ...Ts>
class FixedSizeBlock
{
private:
char block[PPackSizeOf<Ts...>::size];
public:
};
int main( )
{
FixedSizeBlock<char,long> b;
std::cout << sizeof(b) << std::endl;
return 0;
}

I've just found that :
template<typename... List>
inline unsigned int myFunction(const List&... list)
{
return sizeof(std::make_tuple(list...));
}
But :
1) Do I have the guarantee that the result will always be the same on all compilers ?
2) Do the make_tuple will make and overhead at compile-time ?

Related

Generalizing std::conditional_t<>

I have a function that computes a certain object from a given parameter (say, an important node from a graph). Now, when calculating such an object, the function might allocate some memory. Sometimes I want the function to just return the result, and sometimes to return the result plus the memory used to compute it.
I typically solve this binary case like this:
enum class what {
what1, // return, e.g., just an int
what2 // return, e.g., a std::pair<int, std::vector<int>>
};
template <what w>
std::conditional_t<w == what::what1, int, std::pair<int, std::vector<int>>>
calculate_something(const param& p) { ... }
I would like to generalize the solution above to longer enumerations
enum class list_whats {
what1,
what2,
what3,
what4,
what5
};
One possible solution is to nest as many std::conditional_t as needed
template <list_whats what>
std::conditional_t<
what == list_whats::what1,
int,
std::conditional_t<
what == list_whats::what2,
float,
....
>
>
>
calculate_something(const param& p) { ... }
But this is cumbersome and perhaps not too elegant.
Does anyone know how to do this in C++ 17?
EDIT
To make the question perfectly clear: how do I implement the function return_something so as to be able to run the following main?
int main() {
int s1 = return_something<list_whats::what1>();
s1 = 3;
float s2 = return_something<list_whats::what2>();
s2 = 4.0f;
double s3 = return_something<list_whats::what3>();
s3 = 9.0;
std::string s4 = return_something<list_whats::what4>();
s4 = "qwer";
std::vector<int> s5 = return_something<list_whats::what5>();
s5[3] = 25;
}
I don't think you should use std::conditional at all to solve your problem. If I get this right, you want to use a template parameter to tell your function what to return. The elegant way to do this could look something like this:
#include <vector>
enum class what { what1, what2 };
template <what W>
auto compute() {
if constexpr (W == what::what1) {
return 100;
}
if constexpr (W == what::what2) {
return std::pair{100, std::vector<int>{}};
}
}
auto main() -> int {
[[maybe_unused]] const auto as_int = compute<what::what1>();
[[maybe_unused]] const auto as_pair = compute<what::what2>();
}
You can also use template specialization if you prefer another syntax:
template <what W>
auto compute();
template <>
auto compute<what::what1>() {
return 100;
}
template <>
auto compute<what::what2>() {
return std::pair{100, std::vector<int>{}};
}
Here's my approach:
what_pair is a pair that corresponds one enum to one type.
what_type_index accepts a enum and a std::tuple<what_pair<...>...> and searches the tuple map where the enums are equal and returns index. It returns maximum std::size_t value, if no match was found.
what_type is the final type, it is the tuple element at the found position. The program won't compile when the index is std::size_t max value because of invalid std::tuple access.
template<what W, typename T>
struct what_pair {
constexpr static what w = W;
using type = T;
};
template<what w, typename tuple_map>
constexpr auto what_type_index() {
std::size_t index = std::numeric_limits<std::size_t>::max();
auto search_map = [&]<std::size_t... Ints>(std::index_sequence<Ints...>) {
((std::tuple_element_t<Ints, tuple_map>::w == w ? (index = Ints) : 0), ...);
};
search_map(std::make_index_sequence<std::tuple_size_v<tuple_map>>());
return index;
}
template<what w, typename tuple_map>
using what_type = typename
std::tuple_element_t<what_type_index<w, tuple_map>(), tuple_map>::type;
and this is the example usage:
int main() {
using what_map = std::tuple<
what_pair<what::what1, int>,
what_pair<what::what2, float>,
what_pair<what::what3, double>,
what_pair<what::what4, std::string>,
what_pair<what::what5, std::vector<int>>>;
static_assert(std::is_same_v<what_type<what::what1, what_map>, int>);
static_assert(std::is_same_v<what_type<what::what2, what_map>, float>);
static_assert(std::is_same_v<what_type<what::what3, what_map>, double>);
static_assert(std::is_same_v<what_type<what::what4, what_map>, std::string>);
static_assert(std::is_same_v<what_type<what::what5, what_map>, std::vector<int>>);
//compilation error, because 'what6' wasn't specified in the 'what_map'
using error = what_type<what::what6, what_map>;
}
try it out on godbolt.
I found a possible solution that nests two structs: the first takes a list of Boolean values to indicate which type should be used, and the nested struct takes the list of possible types (see conditional_list in the example code below -- the nested structs were inspired by this answer). But perhaps it's not elegant enough. I'm wondering if there is a possible solution of the form
std::conditional_list<
..., // list of Boolean values, (of any length!)
... // list of types (list that should be as long as the first)
>::type
My proposal
#include <type_traits>
#include <iostream>
#include <vector>
// -----------------------------------------------------------------------------
template<auto A, auto... ARGS>
constexpr auto sum = A + sum<ARGS...>;
template<auto A>
constexpr auto sum<A> = A;
// -----------------------------------------------------------------------------
template <bool... values>
static constexpr bool exactly_one_v = sum<values...> == 1;
// -----------------------------------------------------------------------------
template <bool... values>
struct which {
static_assert(exactly_one_v<values...>);
template <std::size_t idx, bool v1, bool... vs>
struct _which_impl {
static constexpr std::size_t value =
(v1 ? idx : _which_impl<idx + 1, vs...>::value);
};
template <std::size_t idx, bool v>
struct _which_impl<idx, v> {
static constexpr std::size_t value = (v ? idx : idx + 1);
};
static constexpr std::size_t value = _which_impl<0, values...>::value;
};
template <bool... conds>
static constexpr std::size_t which_v = which<conds...>::value;
// -----------------------------------------------------------------------------
template <std::size_t ith_idx, typename... Ts>
struct ith_type {
template <std::size_t cur_idx, typename t1, typename... ts>
struct _ith_type_impl {
typedef
std::conditional_t<
ith_idx == cur_idx,
t1,
typename _ith_type_impl<cur_idx + 1, ts...>::type
>
type;
};
template <std::size_t cur_idx, typename t1>
struct _ith_type_impl<cur_idx, t1> {
typedef
std::conditional_t<ith_idx == cur_idx, t1, std::nullptr_t>
type;
};
static_assert(ith_idx < sizeof...(Ts));
typedef typename _ith_type_impl<0, Ts...>::type type;
};
template <std::size_t ith_idx, typename... ts>
using ith_type_t = typename ith_type<ith_idx, ts...>::type;
// -----------------------------------------------------------------------------
template <bool... conds>
struct conditional_list {
template <typename... ts>
struct good_type {
static_assert(sizeof...(conds) == sizeof...(ts));
typedef ith_type_t<which_v<conds...>, ts...> type;
};
};
// -----------------------------------------------------------------------------
enum class list_whats {
what1,
what2,
what3,
what4,
what5,
};
template <list_whats what>
typename conditional_list<
what == list_whats::what1,
what == list_whats::what2,
what == list_whats::what3,
what == list_whats::what4,
what == list_whats::what5
>::template good_type<
int,
float,
double,
std::string,
std::vector<int>
>::type
return_something() noexcept {
if constexpr (what == list_whats::what1) { return 1; }
if constexpr (what == list_whats::what2) { return 2.0f; }
if constexpr (what == list_whats::what3) { return 3.0; }
if constexpr (what == list_whats::what4) { return "42"; }
if constexpr (what == list_whats::what5) { return {1,2,3,4,5}; }
}
int main() {
auto s1 = return_something<list_whats::what1>();
s1 = 3;
auto s2 = return_something<list_whats::what2>();
s2 = 4.0f;
auto s3 = return_something<list_whats::what3>();
s3 = 9.0;
auto s4 = return_something<list_whats::what4>();
s4 = "qwer";
auto s5 = return_something<list_whats::what5>();
s5[3] = 25;
}
An alternative to working only with types is to write a function that returns the specific type, or an identity<type>. It's sometimes more readable. Here is an example:
// if you don't have it in std
template<typename T>
struct identity {
using type = T;
};
enum class what {
what1,
what2,
what3
};
template<what w>
auto return_type_for_calc() {
if constexpr (w == what::what1) {
return identity<int>();
} else if constexpr (w==what::what2) {
return identity<double>();
} else {
return identity<float>();
}
}
template<what w>
decltype(return_type_for_calc<w>())
calculate_something()
{
return {};
}
int main() {
calculate_something<what::what1>();
calculate_something<what::what2>();
return 0;
}
Although I already posted an answer to my own question, and I accepted an answer from another user, I thought I could post another possibility in tackling this problem using the following struct
template <typename... Ts> struct type_sequence { };
which I learnt about in this talk by Andrei Alexandrescu. Since I learnt quite a bit by using it and the result is a bit simpler than the original answer that used two nested structs I thought I would share it here. However, the solution I would actually implement is the one that was accepted.
This is the full code with a main function included. Notice the change in the specification of function return_something. Now this function indicates the return type (which I like very much, perhaps I'm old fashioned) in a more readable way than in my first answer. You can try it out here.
#include <type_traits>
#include <iostream>
#include <vector>
template <bool... values>
struct which {
template <std::size_t idx, bool v1, bool... vs>
struct _which_impl {
static constexpr std::size_t value =
(v1 ? idx : _which_impl<idx + 1, vs...>::value);
};
template <std::size_t idx, bool v>
struct _which_impl<idx, v> {
static constexpr std::size_t value = (v ? idx : idx + 1);
};
static constexpr std::size_t value = _which_impl<0, values...>::value;
};
template <std::size_t ith_idx, typename... Ts>
struct ith_type {
template <std::size_t cur_idx, typename t1, typename... ts>
struct _ith_type_impl {
using type =
std::conditional_t<
ith_idx == cur_idx,
t1,
typename _ith_type_impl<cur_idx + 1, ts...>::type
>;
};
template <std::size_t cur_idx, typename t1>
struct _ith_type_impl<cur_idx, t1> {
using type = std::conditional_t<ith_idx == cur_idx, t1, std::nullptr_t>;
};
using type = typename _ith_type_impl<0, Ts...>::type;
};
template <std::size_t ith_idx, typename... ts>
using ith_type_t = typename ith_type<ith_idx, ts...>::type;
template <bool... conds>
static constexpr std::size_t which_v = which<conds...>::value;
template <typename... Ts> struct type_sequence { };
template <bool... values> struct bool_sequence {
static constexpr std::size_t which = which_v<values...>;
};
template <std::size_t ith_idx, typename... Ts>
struct ith_type<ith_idx, type_sequence<Ts...>> : ith_type<ith_idx, Ts...>
{ };
template <typename bool_sequence, typename type_sequence>
struct conditional_list {
using type = ith_type_t<bool_sequence::which, type_sequence>;
};
template <typename bool_sequence, typename type_sequence>
using conditional_list_t =
typename conditional_list<bool_sequence, type_sequence>::type;
enum class list_whats {
what1,
what2,
what3,
what4,
what5,
};
template <list_whats what>
conditional_list_t<
bool_sequence<
what == list_whats::what1,
what == list_whats::what2,
what == list_whats::what3,
what == list_whats::what4,
what == list_whats::what5
>,
type_sequence<
int,
float,
double,
std::string,
std::vector<int>
>
>
return_something() noexcept {
if constexpr (what == list_whats::what1) { return 1; }
if constexpr (what == list_whats::what2) { return 2.0f; }
if constexpr (what == list_whats::what3) { return 3.0; }
if constexpr (what == list_whats::what4) { return "42"; }
if constexpr (what == list_whats::what5) { return {1,2,3,4,5}; }
}
int main() {
[[maybe_unused]] auto s1 = return_something<list_whats::what1>();
[[maybe_unused]] auto s2 = return_something<list_whats::what2>();
[[maybe_unused]] auto s3 = return_something<list_whats::what3>();
[[maybe_unused]] auto s4 = return_something<list_whats::what4>();
[[maybe_unused]] auto s5 = return_something<list_whats::what5>();
}

Can I allocate a series of variables on the stack based on template arguments?

In a piece of code I'm writing, I receive packets as uint8_t * and std::size_t combination. I can register functions to call with these two parameters, based on which file descriptor the packet was received from. I use an std::map<int, std::function<void(const uint8_t *, std::size_t)> > handlers to keep track of which function to call.
I would like to be able to (indirectly) register functions with arbitrary arguments. I already have a function like this to transform from the uint8_t * and std::size_t to separate variables:
int unpack(const uint8_t *buf, std::size_t len) { return 0; }
template <typename T, typename... Types>
int unpack(const uint8_t *buf, std::size_t len, T &var1, Types... var2) {
static_assert(std::is_trivially_copyable<T>::value, "unpack() only works for primitive types");
if (len < sizeof(T)) return -1;
var1 = *reinterpret_cast<const T *>(buf);
const auto sum = unpack(buf + sizeof(T), len - sizeof(T), var2...);
const auto ret = (sum == -1) ? -1 : sum + sizeof(T);
return ret;
}
My question is: Is it possible with C++20 to auto-generate a function that convers from uint8_t * and std::size_t to the arguments that a passed function needs?
I would like to be able to do this:
void handler(unsigned int i) { ... }
int main(int argc, char ** argv) {
/* some code generating an fd */
handlers[fd] = function_returning_an_unpacker_function_that_calls_handler(handler);
edit: I realize I went a bit too short on my answer, as some mentioned (thanks!).
I am wondering if it is possible (and if so, how?) to implement the function_returning_an_unpacker_function_that_calls_handler function. I started out doing something like this (written from memory):
template<typename... Types>
std::function<void(const uint8_t * buf, std::size_t)>
function_returning_an_unpacker_function_that_calls_handler(std::function<void(Types...)> function_to_call) {
const auto ret = new auto([fun](const uint8_t * buf, std::size_t len) -> void {
const auto unpack_result = unpack(buf, len, list_of_variables_based_on_template_params);
if(unpack_result == -1) return nullptr;
function_to_call(list_of_variables_based_on_template_params);
};
return ret;
}
This is also why I supplied the unpack function. The problem I'm encountering is that I'm struggling with the list_of_variables_based_on_template_params bit. I haven't found any way to generate a list of variables that I can repeat identically in two places.
I also looked a little bit into using std::tuple::tie and friends, but I didn't see a solution there either.
It's possible, just annoying to write.
First you need a trait to get parameters from a function type:
template <typename T>
struct FuncTraits {};
#define GEN_FUNC_TRAITS_A(c, v, ref, noex) \
template <typename R, typename ...P> \
struct FuncTraits<R(P...) c v ref noex> \
{ \
template <template <typename...> typename T> \
using ApplyParams = T<P...>; \
};
#define GEN_FUNC_TRAITS_B(c, v, ref) \
GEN_FUNC_TRAITS_A(c, v, ref,) \
GEN_FUNC_TRAITS_A(c, v, ref, noexcept)
#define GEN_FUNC_TRAITS_C(c, v) \
GEN_FUNC_TRAITS_B(c, v,) \
GEN_FUNC_TRAITS_B(c, v, &) \
GEN_FUNC_TRAITS_B(c, v, &&)
#define GEN_FUNC_TRAITS(c) \
GEN_FUNC_TRAITS_C(c,) \
GEN_FUNC_TRAITS_C(c, volatile)
GEN_FUNC_TRAITS()
GEN_FUNC_TRAITS(const)
Then some templates to analyze what kind of callable (function, function pointer, or a functor) you got, and apply the trait accordingly:
template <typename T> struct RemoveMemPtr {using type = T;};
template <typename T, typename C> struct RemoveMemPtr<T C::*> {using type = T;};
template <typename T>
struct ToFuncType {};
template <typename T>
requires std::is_function_v<std::remove_pointer_t<T>>
struct ToFuncType<T> {using type = std::remove_pointer_t<T>;};
template <typename T>
requires requires {&T::operator();}
struct ToFuncType<T>
{
using type = typename RemoveMemPtr<decltype(&T::operator())>::type;
};
Then you can make a templated functor that automatically unwraps the arguments. Since Unwrap() must be called in order, and function arguments are evaluated in unspecified order, we need a tuple (or something similar) that accepts a braced list:
template <typename T>
T Unpack(char *&, std::size_t &)
{
std::cout << __PRETTY_FUNCTION__ << '\n';
return {};
}
template <typename F>
struct WrapFunctor
{
template <typename ...P>
struct Impl
{
std::decay_t<F> func;
void operator()(char *p, std::size_t n)
{
std::apply(func, std::tuple{Unpack<P>(p, n)...});
}
};
};
template <typename F>
auto Wrap(F &&func)
{
using Functor = typename FuncTraits<typename ToFuncType<std::remove_cvref_t<F>>::type>::template ApplyParams<WrapFunctor<F>::template Impl>;
return Functor{std::forward<F>(func)};
}
Finally, some tests:
void foo(int, float, char)
{
std::cout << __PRETTY_FUNCTION__ << '\n';
}
int main()
{
Wrap(foo)(nullptr, 42);
Wrap(&foo)(nullptr, 42);
Wrap([](int, float, char){std::cout << __PRETTY_FUNCTION__ << '\n';})(nullptr, 42);
}
I've changed the signature of Unpack() to take the parameters by reference and unpack one variable at a time.
Almost forgot: var1 = *reinterpret_cast<const T *>(buf); is a strict aliasing violation and UB. Prefer memcpy.
This answer is very similar to the first one, but it leverages the use of CTAD and std::function to figure out the function signature.
Creates a tuple based on the function signature, and passes both the argument types and the elements from the tuple on to unpack.
#include <iostream>
#include <tuple>
#include <type_traits>
#include <cstring>
#include <functional>
int unpack(const uint8_t *buf, std::size_t len) { return 0; }
template <typename T, typename... Types>
int unpack(const uint8_t *buf, std::size_t len, T &var1, Types&... var2) {
static_assert(std::is_trivially_copyable<T>::value, "unpack() only works for primitive types");
if (len < sizeof(T)) return -1;
var1 = *reinterpret_cast<const T *>(buf);
std::cout << "In unpack " << var1 << "\n";
const auto sum = unpack(buf + sizeof(T), len - sizeof(T), var2...);
const auto ret = (sum == -1) ? -1 : sum + sizeof(T);
return ret;
}
template<typename T, typename R, typename... Args>
std::function<void(const uint8_t * buf, std::size_t)>
unpack_wrapper_impl(T function_to_call, std::function<R(Args...)>) {
return [function_to_call](const uint8_t *buf, std::size_t len) -> void {
std::tuple<std::decay_t<Args>...> tup;
std::apply([&](auto&... args) {
unpack(buf, len, args...);
}, tup);
std::apply(function_to_call, tup);
};
}
template<typename T>
std::function<void(const uint8_t * buf, std::size_t)>
unpack_wrapper(T&& function_to_call) {
return unpack_wrapper_impl(std::forward<T>(function_to_call), std::function{function_to_call});
}
void test(int a, int b) {
std::cout << a << " " << b << "\n";
}
int main() {
int a= 5, b = 9;
uint8_t* buf = new uint8_t[8];
std::memcpy(buf, &a, 4);
std::memcpy(buf + 4, &b, 4);
auto f = unpack_wrapper(test);
f(buf, 8);
}

Compile-time C++ function to check whether all template argument types are unique

There is a nice question (Which substitution failures are not allowed in requires clauses?) proposing the next problem.
One needs to write a compile-time function template<typename... Ts> constexpr bool allTypesUnique() that will return true if all argument types are unique, and false otherwise. And the restriction is not to compare the argument types pairwise. Unfortunately, the answer only explains why such function cannot be implemented with some particular approach.
I think the solution can be achieved using multiple inheritance. The idea is to make a class inherited from a number of classes: one for each type T in Ts. And each such class defines a virtual function with a signature depending on T. If some T is found more than once in Ts then function f in a child class will override the function in a base class and it can be detected:
template<typename> struct A{};
template<typename T, typename... Ts>
struct B : B<Ts...> {
using B<Ts...>::f;
constexpr virtual void f(A<T>, bool & unique) { if( ++count > 1 ) unique = false; }
int count = 0;
};
template<typename T>
struct B<T> {
constexpr virtual void f(A<T>, bool & unique) { if( ++count > 1 ) unique = false; }
int count = 0;
};
template<typename... Ts>
constexpr bool allTypesUnique() {
B<Ts...> b;
bool res = true;
( b.f( A<Ts>{}, res ), ... );
return res;
}
int main() {
static_assert( allTypesUnique<void>() );
static_assert( allTypesUnique<void, int&>() );
static_assert( !allTypesUnique<int&, int&>() );
static_assert( allTypesUnique<char, short, int>() );
static_assert( !allTypesUnique<char, short, char>() );
}
Demo: https://gcc.godbolt.org/z/8jhnE7P11
Just curious, is the solution correct and is there a simpler solution for this problem?
A simpler solution can be obtained in compilers supporting class size optimization via C++20 attribute [[no_unique_address]] for empty members. If all class empty members have distinct types, then its sizeof will be 1. If some member types are repeating then they cannot share the same address and sizeof will be more than 1.
Solution code:
template<typename> struct A{};
template<typename T, typename... Ts>
struct B : B<Ts...> {
[[no_unique_address]] A<T> a;
};
template<typename T>
struct B<T> {
[[no_unique_address]] A<T> a;
};
template<typename... Ts>
constexpr bool allTypesUnique() {
if constexpr (sizeof...(Ts) <= 1 )
return true;
else
return sizeof(B<Ts...>) == 1;
}
int main() {
static_assert( allTypesUnique<void>() );
static_assert( allTypesUnique<void, int&>() );
static_assert( !allTypesUnique<int&, int&>() );
static_assert( allTypesUnique<char, short, int>() );
static_assert( !allTypesUnique<char, short, char>() );
}
Demo: https://gcc.godbolt.org/z/577EP1774
A single constexpr function can be constructed to answer this question:
template <class T, class... Ts>
constexpr bool unique_types()
{
if constexpr (sizeof...(Ts)) {
return !(std::is_same_v<T,Ts> || ...) && unique_types<Ts...>();
}
return true;
}
Demo
the code above is the compile time equivalent of
for (int i(0); i < n; ++i)
for (int j(i + 1); j < n; ++j)
{ /* check type[i] vs type[j] */ }
If you want any variation in comparison (e.g. consider int same as int&) just modify the std::is_same_v (e.g. std::is_same_v<std::decay_t<T>, std::decay_t<Ts>>)
If you use virtual base classes depending on each of the given types, you will get exact one base class instance for every unique type in the resulting class. If the number of given types is the number of generated base classes, each type was unique. You can "measure" the number of generated base classes by its size but must take care that you have a vtable pointer inside which size is implementation dependent. As this, each generated type should be big enough to hide alignment problems.
BTW: It works also for reference types.
template < typename T> struct AnyT { char i[128]; };
template < typename FIRST, typename ... T>
struct CheckT: virtual AnyT<FIRST>, virtual CheckT<T...> { };
template < typename FIRST >
struct CheckT<FIRST>: virtual AnyT<FIRST> {};
template < typename ... T>
constexpr bool allTypesUnique()
{
using T1 = CheckT<int>;
using T2 = CheckT<bool, int>;
constexpr std::size_t s1 = sizeof( T1 );
constexpr std::size_t s2 = sizeof( T2 );
constexpr std::size_t diff = s2 - s1;
constexpr std::size_t base = s1 - diff;
constexpr std::size_t measure = sizeof( CheckT< T...> );
return !((sizeof...(T)*diff+base) - measure);
}
int main() {
static_assert( allTypesUnique<void>() );
static_assert( allTypesUnique<void, int>() );
static_assert( !allTypesUnique<void, void>() );
static_assert( allTypesUnique<char, short, int>() );
static_assert( !allTypesUnique<char, short, char>() );
}
Demo
I think you can try to elaborate on using compile-time type ids.
Here is a partly implemented C++14 version (compile-time sorting is required):
#include <cstddef>
#include <type_traits>
#include <utility>
namespace detail {
template<typename T>
struct type_id {
constexpr static int value{};
};
template<const int *const A, const int *const B>
struct same_address : std::false_type {};
template<const int *const A>
struct same_address<A, A> : std::true_type {};
}// namespace detail
// TODO: implement
template<const int *const... I>
struct sort_array {
using type = std::integer_sequence<const int *const, I...>;
};
template<typename>
struct find_duplicates;
template<const int *const A, const int *const B, const int *const... I>
struct find_duplicates<std::integer_sequence<const int *const, A, B, I...>> {
constexpr static bool value = std::conditional_t<detail::same_address<A, B>::value,
std::true_type,
find_duplicates<std::integer_sequence<const int *const, B, I...>>>::value;
};
template<>
struct find_duplicates<std::integer_sequence<const int *const>> {
constexpr static bool value = false;
};
template<const int *const I>
struct find_duplicates<std::integer_sequence<const int *const, I>> {
constexpr static bool value = false;
};
template<typename... T>
constexpr bool all_types_unique() {
return !find_duplicates<typename sort_array<&detail::type_id<T>::value...>::type>::value;
};
int main() {
static_assert(detail::same_address<&detail::type_id<int>::value,
&detail::type_id<int>::value>::value,
"");
static_assert(!detail::same_address<&detail::type_id<int>::value,
&detail::type_id<double>::value>::value,
"");
static_assert(all_types_unique<>(), "");
static_assert(all_types_unique<int>(), "");
static_assert(all_types_unique<int, double>(), "");
static_assert(all_types_unique<int, double, char>(), "");
static_assert(!all_types_unique<int, int>(), "");
static_assert(!all_types_unique<int, int, int>(), "");
return 0;
}
https://godbolt.org/z/E4G6YchE5

Determine member offset of struct or tuple in template

I want to write a template function that writes tables to HDF5 files.
The signature should look similar to
template<typename record> void writeTable(const std::vector<record>& data);
where record is a struct, or
template<typename... elements>
void writeTable(const std::vector<std::tuple<elements...>>& data);
The actual implementation would have more parameters to determine the destionation, etc.
To write the data I need to define a HDF5 compound type, which contains the name and the offset of the members. Usually you would use the HOFFSET macro the get the field offset, but as I don't know the struct fields beforehand I can't do that.
What I tried so far was constructing a struct type from the typename pack. The naive implementation did not have standard layout, but the implementation here does. All that's left is get the offsets of the members. I would like to expand the parameter pack into an initializer list with the offsets:
#include <vector>
template<typename... members> struct record {};
template<typename member, typename... members> struct record<member, members...> :
record<members...> {
record(member m, members... ms) : record<members...>(ms...), tail(m) {}
member tail;
};
template<typename... Args> void
make_table(const std::string& name, const std::vector<record<Args...>>& data) {
using record_type = record<Args...>;
std::vector<size_t> offsets = { get_offset(record_type,Args)... };
}
int main() {
std::vector<record<int, float>> table = { {1, 1.0}, {2, 2.0} };
make_table("table", table);
}
Is there a possible implementation for get_offset? I would think not, because in the case of record<int, int> it would be ambiguous. Is there another way to do it?
Or is there any other way I could approach this problem?
Calculating offsets is quite simple. Given a tuple with types T0, T1 ... TN. The offset of T0 is 0 (as long as you use alignas(T0) on your char array. The offset of T1 is the sizeof(T0) rounded up to alignof(T1).
In general, the offset of TB (which comes after TA) is round_up(offset_of<TA>() + sizeof(TA), alignof(TB)).
Calculating the offsets of elements in a std::tuple could be done like this:
constexpr size_t roundup(size_t num, size_t multiple) {
const size_t mod = num % multiple;
return mod == 0 ? num : num + multiple - mod;
}
template <size_t I, typename Tuple>
struct offset_of {
static constexpr size_t value = roundup(
offset_of<I - 1, Tuple>::value + sizeof(std::tuple_element_t<I - 1, Tuple>),
alignof(std::tuple_element_t<I, Tuple>)
);
};
template <typename Tuple>
struct offset_of<0, Tuple> {
static constexpr size_t value = 0;
};
template <size_t I, typename Tuple>
constexpr size_t offset_of_v = offset_of<I, Tuple>::value;
Here's a test suite. As you can see from the first test, the alignment of elements is taken into account.
static_assert(offset_of_v<1, std::tuple<char, long double>> == 16);
static_assert(offset_of_v<2, std::tuple<char, char, long double>> == 16);
static_assert(offset_of_v<3, std::tuple<char, char, char, long double>> == 16);
static_assert(offset_of_v<4, std::tuple<char, char, char, char, long double>> == 16);
static_assert(offset_of_v<0, std::tuple<int, double, int, char, short, long double>> == 0);
static_assert(offset_of_v<1, std::tuple<int, double, int, char, short, long double>> == 8);
static_assert(offset_of_v<2, std::tuple<int, double, int, char, short, long double>> == 16);
static_assert(offset_of_v<3, std::tuple<int, double, int, char, short, long double>> == 20);
static_assert(offset_of_v<4, std::tuple<int, double, int, char, short, long double>> == 22);
static_assert(offset_of_v<5, std::tuple<int, double, int, char, short, long double>> == 32);
I hardcoded the offsets in the above tests. The offsets are correct if the following tests succeed.
static_assert(sizeof(char) == 1 && alignof(char) == 1);
static_assert(sizeof(short) == 2 && alignof(short) == 2);
static_assert(sizeof(int) == 4 && alignof(int) == 4);
static_assert(sizeof(double) == 8 && alignof(double) == 8);
static_assert(sizeof(long double) == 16 && alignof(long double) == 16);
std::tuple seems to store it's elements sequentially (without sorting them to optimize padding). That's proven by the following tests. I don't think the standard requires std::tuple to be implemented this way so I don't think the following tests are guaranteed to succeed.
template <size_t I, typename Tuple>
size_t real_offset(const Tuple &tup) {
const char *base = reinterpret_cast<const char *>(&tup);
return reinterpret_cast<const char *>(&std::get<I>(tup)) - base;
}
int main(int argc, char **argv) {
using Tuple = std::tuple<int, double, int, char, short, long double>;
Tuple tup;
assert((offset_of_v<0, Tuple> == real_offset<0>(tup)));
assert((offset_of_v<1, Tuple> == real_offset<1>(tup)));
assert((offset_of_v<2, Tuple> == real_offset<2>(tup)));
assert((offset_of_v<3, Tuple> == real_offset<3>(tup)));
assert((offset_of_v<4, Tuple> == real_offset<4>(tup)));
assert((offset_of_v<5, Tuple> == real_offset<5>(tup)));
}
Now that I've gone to all of this effort, would that real_offset function suit your needs?
This is a minimal implementation of a tuple that accesses a char[] with offset_of. This is undefined behavior though because of the reinterpret_cast. Even though I'm constructing the object in the same bytes and accessing the object in the same bytes, it's still UB. See this answer for all the standardese. It will work on every compiler you can find but it's UB so just use it anyway. This tuple is standard layout (unlike std::tuple). If the elements of your tuple are all trivially copyable, you can remove the copy and move constructors and replace them with memcpy.
template <typename... Elems>
class tuple;
template <size_t I, typename Tuple>
struct tuple_element;
template <size_t I, typename... Elems>
struct tuple_element<I, tuple<Elems...>> {
using type = std::tuple_element_t<I, std::tuple<Elems...>>;
};
template <size_t I, typename Tuple>
using tuple_element_t = typename tuple_element<I, Tuple>::type;
template <typename Tuple>
struct tuple_size;
template <typename... Elems>
struct tuple_size<tuple<Elems...>> {
static constexpr size_t value = sizeof...(Elems);
};
template <typename Tuple>
constexpr size_t tuple_size_v = tuple_size<Tuple>::value;
constexpr size_t roundup(size_t num, size_t multiple) {
const size_t mod = num % multiple;
return mod == 0 ? num : num + multiple - mod;
}
template <size_t I, typename Tuple>
struct offset_of {
static constexpr size_t value = roundup(
offset_of<I - 1, Tuple>::value + sizeof(tuple_element_t<I - 1, Tuple>),
alignof(tuple_element_t<I, Tuple>)
);
};
template <typename Tuple>
struct offset_of<0, Tuple> {
static constexpr size_t value = 0;
};
template <size_t I, typename Tuple>
constexpr size_t offset_of_v = offset_of<I, Tuple>::value;
template <size_t I, typename Tuple>
auto &get(Tuple &tuple) noexcept {
return *reinterpret_cast<tuple_element_t<I, Tuple> *>(tuple.template addr<I>());
}
template <size_t I, typename Tuple>
const auto &get(const Tuple &tuple) noexcept {
return *reinterpret_cast<tuple_element_t<I, Tuple> *>(tuple.template addr<I>());
}
template <typename... Elems>
class tuple {
alignas(tuple_element_t<0, tuple>) char storage[offset_of_v<sizeof...(Elems), tuple<Elems..., char>>];
using idx_seq = std::make_index_sequence<sizeof...(Elems)>;
template <size_t I>
void *addr() {
return static_cast<void *>(&storage + offset_of_v<I, tuple>);
}
template <size_t I, typename Tuple>
friend auto &get(const Tuple &) noexcept;
template <size_t I, typename Tuple>
friend const auto &get(Tuple &) noexcept;
template <size_t... I>
void default_construct(std::index_sequence<I...>) {
(new (addr<I>()) Elems{}, ...);
}
template <size_t... I>
void destroy(std::index_sequence<I...>) {
(get<I>(*this).~Elems(), ...);
}
template <size_t... I>
void move_construct(tuple &&other, std::index_sequence<I...>) {
(new (addr<I>()) Elems{std::move(get<I>(other))}, ...);
}
template <size_t... I>
void copy_construct(const tuple &other, std::index_sequence<I...>) {
(new (addr<I>()) Elems{get<I>(other)}, ...);
}
template <size_t... I>
void move_assign(tuple &&other, std::index_sequence<I...>) {
(static_cast<void>(get<I>(*this) = std::move(get<I>(other))), ...);
}
template <size_t... I>
void copy_assign(const tuple &other, std::index_sequence<I...>) {
(static_cast<void>(get<I>(*this) = get<I>(other)), ...);
}
public:
tuple() noexcept((std::is_nothrow_default_constructible_v<Elems> && ...)) {
default_construct(idx_seq{});
}
~tuple() {
destroy(idx_seq{});
}
tuple(tuple &&other) noexcept((std::is_nothrow_move_constructible_v<Elems> && ...)) {
move_construct(other, idx_seq{});
}
tuple(const tuple &other) noexcept((std::is_nothrow_copy_constructible_v<Elems> && ...)) {
copy_construct(other, idx_seq{});
}
tuple &operator=(tuple &&other) noexcept((std::is_nothrow_move_assignable_v<Elems> && ...)) {
move_assign(other, idx_seq{});
return *this;
}
tuple &operator=(const tuple &other) noexcept((std::is_nothrow_copy_assignable_v<Elems> && ...)) {
copy_assign(other, idx_seq{});
return *this;
}
};
Alternatively, you could use this function:
template <size_t I, typename Tuple>
size_t member_offset() {
return reinterpret_cast<size_t>(&std::get<I>(*static_cast<Tuple *>(nullptr)));
}
template <typename Member, typename Class>
size_t member_offset(Member (Class::*ptr)) {
return reinterpret_cast<size_t>(&(static_cast<Class *>(nullptr)->*ptr));
}
template <auto MemPtr>
size_t member_offset() {
return member_offset(MemPtr);
}
Once again, this is undefined behavior (because of the nullptr dereference and the reinterpret_cast) but it will work as expected with every major compiler. The function cannot be constexpr (even though member offset is a compile-time calculation).
Not sure to understand what do you exactly want but... what about using recursion based on a index sequence (starting from C++14) something as follows?
#include <vector>
#include <utility>
#include <iostream>
template <typename... members>
struct record
{ };
template <typename member, typename... members>
struct record<member, members...> : record<members...>
{
record (member m, members... ms) : record<members...>(ms...), tail(m)
{ }
member tail;
};
template <std::size_t, typename, std::size_t = 0u>
struct get_offset;
template <std::size_t N, typename A0, typename ... As, std::size_t Off>
struct get_offset<N, record<A0, As...>, Off>
: public get_offset<N-1u, record<As...>, Off+sizeof(A0)>
{ };
template <typename A0, typename ... As, std::size_t Off>
struct get_offset<0u, record<A0, As...>, Off>
: public std::integral_constant<std::size_t, Off>
{ };
template <typename... Args, std::size_t ... Is>
auto make_table_helper (std::string const & name,
std::vector<record<Args...>> const & data,
std::index_sequence<Is...> const &)
{ return std::vector<std::size_t>{ get_offset<Is, record<Args...>>::value... }; }
template <typename... Args>
auto make_table (std::string const & name,
std::vector<record<Args...>> const & data)
{ return make_table_helper(name, data, std::index_sequence_for<Args...>{}); }
int main ()
{
std::vector<record<int, float>> table = { {1, 1.0}, {2, 2.0} };
auto v = make_table("table", table);
for ( auto const & o : v )
std::cout << o << ' ';
std::cout << std::endl;
}
Unfortunately isn't an efficient solution because the last value is calculated n-times.

SFINAE: function templates optimation

I would like iterate over an tuple in some way with member function templates (for later create a new type of tuple from the given template type T).
However, the break condition (function) is not used so I get this error:
invalid use of incomplete type: 'class std::tuple_element<0ul, std::tuple<> >'
The problem seems to be, that even though N == size of the tuple, std::tuple_element_t is evaluated for N != size and not handled as SFINAE.
Both examples showing different not working solutions. What do I wrong?
Note: The function for evaluated with is_same is omitted to minimize the example.
#include <type_traits>
#include <tuple>
template<typename...Ts>
struct A
{
using tuple = std::tuple<Ts...>;
static constexpr std::size_t size = sizeof...(Ts);
template<typename T, std::size_t N = 0, typename std::enable_if_t<N == size>* = nullptr>
int get()
{
return 0;
}
template<typename T, std::size_t N = 0, typename std::enable_if_t<N != size && !std::is_same<T, std::tuple_element_t<N, tuple>>::value>* = nullptr>
int get()
{
return get<T, N + 1>() - 1;
}
};
int main()
{
A<int, float, double, float, float> a;
return a.get<char>();
}
Live Example 1
#include <type_traits>
#include <tuple>
template<typename...Ts>
struct A
{
using tuple = std::tuple<Ts...>;
static constexpr std::size_t size = sizeof...(Ts);
template<typename T, std::size_t N = 0>
std::enable_if_t<N == size, int> get()
{
return 0;
}
template<typename T, std::size_t N = 0>
std::enable_if_t<N != size && !std::is_same<T, std::tuple_element_t<N, tuple>>::value, int> get()
{
return get<T, N + 1>() - 1;
}
};
int main()
{
A<int, float, double, float, float> a;
return a.get<char>();
}
Live Example 2
One workaround would be to use a third function to evaluate until sizeof tuple - 2 and than evaluate sizeof tuple - 1, but Is this really necessary?
#include <type_traits>
#include <tuple>
template<typename...Ts>
struct A
{
using tuple = std::tuple<Ts...>;
static constexpr std::size_t size = sizeof...(Ts);
template<typename T, std::size_t N = 0, typename std::enable_if_t<(N == size - 1) && std::is_same<T, std::tuple_element_t<N, tuple>>::value>* = nullptr>
int get()
{
return 1;
}
template<typename T, std::size_t N = 0, typename std::enable_if_t<(N == size - 1) && !std::is_same<T, std::tuple_element_t<N, tuple>>::value>* = nullptr>
int get()
{
return 2;
}
template<typename T, std::size_t N = 0, typename std::enable_if_t<(N < size - 1) && !std::is_same<T, std::tuple_element_t<N, tuple>>::value>* = nullptr>
int get()
{
return get<T, N + 1>() - 1;
}
};
int main()
{
A<int, float, double, float, float> a;
return a.get<char>();
}
Live Example 3
As suggested by #PiotrSkotnicki in the comments to the question, here is your second example once fixed:
#include <type_traits>
#include <tuple>
template<typename...Ts>
struct A
{
using tuple = std::tuple<Ts...>;
static constexpr std::size_t size = sizeof...(Ts);
template<typename T, std::size_t N = 0>
std::enable_if_t<N == size-1, int>
get()
{
return std::is_same<T, std::tuple_element_t<N, tuple>>::value ? N : 0;
}
template<typename T, std::size_t N = 0>
std::enable_if_t<N != size-1 && !std::is_same<T, std::tuple_element_t<N, tuple>>::value, int>
get()
{
return get<T, N + 1>() - 1;
}
};
int main()
{
A<int, float, double, float, float> a;
return a.get<char>();
}
What was the problem?
Consider the following line:
std::enable_if_t<N != size && !std::is_same<T, std::tuple_element_t<N, tuple>>::value, int> get()
In this case, N was substituted in order to evaluate the condition of the enable_if, even when N == size (substitution is mandatory to find that N == size indeed).
Thus, the tuple_element_t (let me say) issued an out of range and that's why you got the compilation error.
I've simply updated your code to avoid reaching size while iterating over N. It was a matter of using size-1 as a value on which to switch between functions.
In a comment to this answer the OP said:
It does solve the problem but not for automatic type return type deduction based on which function is used (returning int was just an example). I should have been clearer on this.
It follows a minimal, working example that probably solves the problem also for that.
It's far easier to reason in terms of inheritance and tag dispatching in this case, so as to reduce the boilerplate due to sfinae. Moreover, one can use specializations to introduce specific behaviors for specific types if needed.
The final case, the one for the type that is not part of the types list, is easily handled in a dedicated function as well.
It follows the code:
#include <type_traits>
#include <tuple>
template<typename>
struct tag {};
template<typename...>
struct B;
template<typename T, typename... Ts>
struct B<T, Ts...>: B<Ts...> {
using B<Ts...>::get;
auto get(tag<T>) {
return T{};
}
};
template<>
struct B<> {
template<typename T>
auto get(tag<T>) {
return nullptr;
}
};
template<typename...Ts>
struct A: private B<Ts...>
{
template<typename T>
auto get() {
return B<Ts...>::get(tag<T>{});
}
};
int main()
{
A<int, float, double, float, float> a;
static_assert(std::is_same<decltype(a.get<char>()), std::nullptr_t>::value, "!");
static_assert(std::is_same<decltype(a.get<float>()), float>::value, "!");
}
What about using an additional struct that, with partial specialization, can avoid the use of std::tuple_element_t ?
I mean, something like
template <typename T, std::size_t N>
struct checkType
{ constexpr static bool value
= std::is_same<T, std::tuple_element_t<N, tuple>>::value; };
template <typename T>
struct checkType<T, size>
{ constexpr static bool value = false; };
template <typename, std::size_t N = 0>
std::enable_if_t<N == size, int> get ()
{ return 0; }
template <typename T, std::size_t N = 0>
std::enable_if_t<(N < size) && ! checkType<T, N>::value, int> get()
{ return get<T, N + 1>() - 1; }