Hi I was trying to implement a C++ concept-like feature (C++14) in C++11. The idea is just to write the wrapper function for std::for_each() algorithm where I just check whether the 3rd argument is a function or not. So I wrote the following code, however I am not able to compile it as it should be. I am using Ubuntu12.04 with gcc4.8.1.
test_1.cpp
#include<type_traits>
#include<iostream>
#include<vector>
#include<algorithm>
void display(int& val) {
std::cout <<val<<std::endl;
}
template<typename T>
constexpr bool Is_valid_function(T& a) {
return std::is_function<T>::value;
}
template<typename T>
void check_valid_function(T& a) {
static_assert(Is_valid_function(a), "Not The Valid Function");
}
template <class InputIterator, class Function>
Function my_for_each(InputIterator first, InputIterator last, Function f) {
/* Concept Like Implementation to just check whether f is function or not */
check_valid_function(f);
return for_each(first, last, f) ;
}
void learn_static_assert_and_typetraits(void)
{
std::vector<int> vec_x{1,2,3,4,5};
my_for_each(vec_x.begin(), vec_x.end(), display);
}
int main(int argc, const char* argv[]) {
learn_static_assert_and_typetraits();
return 0;
}
I am getting the following compilation error from which I can see that the static_assert() fails which is not correct as display is valid function.
Compilation Error
test_3.cpp: In instantiation of ‘void check_valid_function(T&) [with T = void (*)(int&)]’:
test_3.cpp:27:26: required from ‘Function my_for_each(InputIterator, InputIterator, Function) [with InputIterator = __gnu_cxx::__normal_iterator >; Function = void (*)(int&)]’
test_3.cpp:35:50: required from here
test_3.cpp:19:3: error: static assertion failed: Not The Valid Function
static_assert(Is_valid_function(a), "Not The Valid Function");
^
However if I do the same thing for the other type_traits function, I am getting the following error which is correct and expected.
test_2.cpp
#include<type_traits>
template<typename T>
constexpr bool Is_floating_point(T& a) {
return std::is_floating_point<T>::value;
}
template<typename T>
void f(T& a) {
static_assert(Is_floating_point(a), "Non-Float Type Data");
}
void learn_static_assert_and_typetraits(void) {
float y{10.0};
f(y);
int x{100};
f(x);
}
int main(int argc, const char* argv[]) {
learn_static_assert_and_typetraits();
return 0;
}
Compiler Output
test_2.cpp: In instantiation of ‘void f(T&) [with T = int]’:
test_2.cpp:19:6: required from here
test_2.cpp:11:3: error: static assertion failed: Non-Float Type Data
static_assert(Is_floating_point(a), "Non-Float Type Data");
^
Question
So, I wanted to understand why my first program is not working as it should be, whether there is bug in my code/understanding or it is something else. I hope the above data would be sufficient to understand my question. However if anyone wants some additional data, please let me know.
The issue is here:
template <class InputIterator, class Function>
Function my_for_each(InputIterator first, InputIterator last, Function f)
invoked via:
my_for_each(vec_x.begin(), vec_x.end(), display);
This deduces Function (of my_for_each) to be a function pointer; for
void display(int& val)
the deduced type is void(*)(int&). The type trait std::is_function however checks if the passed type is a function type, not a function pointer type.
One solution is to remove the pointer:
template<typename T>
constexpr bool Is_valid_function(T& a) {
return std::is_function<typename std::remove_pointer<T>::type>::value;
}
But, as clang++ reveals, this still isn't sufficient:
template<typename T>
void check_valid_function(T& a) {
static_assert(Is_valid_function(a), "Not The Valid Function");
}
a as a function parameter (even if check_valid_function was constexpr!) is not a compile-time constant, therefore it may not appear in a constant expression (inside the function body). Hence, Is_valid_function(a) may not appear as the check for the static_assert. It might be possible to use something similar to declval, e.g.
static_assert(Is_valid_function( declval<T&>() ), "Not The Valid Function");
but unfortunately, declval is not constexpr, and I don't know how to write a constexpr version. So, you could pass a pointer instead:
static_assert(Is_valid_function(static_cast<T*>(nullptr)),
"Not a valid function");
For this, you need to rewrite Is_valid_function as follows:
template<typename T>
constexpr bool Is_valid_function(T*) {
return std::is_function<typename std::remove_pointer<T>::type>::value;
}
Note: the passed argument here is a pointer to a pointer to a function, but the parameter T* deduced T to be a pointer to a function, as before (hence the change in the signature). You might want to reflect that in the function name, if you choose this solution.
Other issues:
Relying on ADL for Standard Library algorithms
return for_each(first, last, f) ;
As far as I can see, this relies on ADL. But the iterators (and the function) are not required to be in namespace std (even for vector::iterator etc.), so you shouldn't rely on ADL:
return std::for_each(first, last, f);
Use of non-const refs for functions that don't need to modify their arguments, e.g.
constexpr bool Is_valid_function(T& a)
If you don't need to modify an argument, you should either pass it by value or by const reference, e.g.
constexpr bool Is_valid_function(T const& a)
"Wrong check" If this code is just for educational purposes, this isn't an issue. However, the check if the passed argument is of a function type is the "wrong check" when trying to check if the argument valid for a Standard Library algorithm. You should rather check whether f(*first) is well-formed. This allows for function objects and checks if the argument type is "valid".
Related
I'm playing around with some C++11 features, and I encountered the following:
#include <iostream>
#include <vector>
template <class T>
void map(std::vector<T>& values, T(*func)(T)) {
for (int &value : values) {
value = func(value);
}
}
int mul2(int x) {
return 2*x;
}
auto mul3 = [](int value) {
return value * 3;
};
int main() {
std::vector<int> v = { 1,2,3,4,5 };
map(v, mul3);
for (auto value : v) {
std::cout << value << std::endl;
}
}
using map with mul2 works as expected, but when I use the mul3 function it gives a compilation error. I expected that auto in this case would give me a int function pointer, but it seems that is not the case here. Anybody could explain this behaviour?
The lambda can implicitly be converted to a function pointer, but that's not what's failing here. Rather, the compiler is failing to deduce T because the lambda-to-function-pointer conversion doesn't happen during deduction.
main.cpp:5:6: note: template argument deduction/substitution failed:
main.cpp:21:16: note: mismatched types 'T (*)(T)' and '<lambda(int)>'
21 | map(v, mul3);
| ^
The compiler can make the connection between T(*)(T) and int(*)(int) and it can make the connection between int(*)(int) and the lambda type, but it can't make the connection between T(*)(T) and the lambda type.
You can fix this by making one of the two connections for it: explicitly specifying the function's template argument, or casting the lambda to the function pointer type. The first skips the deduction step an then the implicit lambda-to-function-pointer conversion succeeds. The second allows deduction to succeed because the second parameter is a compatible function pointer type.
// Option 1: Specifying T allows implicit conversion of the lambda to fnptr
map<int>(v, mul3);
// Option 2a: Cast of lambda to fnptr allows T to be deduced
map(v, static_cast<int(*)(int)>(mul3));
// Option 2b: Converting lambda to fnptr at mul3 initialization allows T to be deduced
int (*mul3)(int) = [](int value) {
return value * 3;
};
However, I would recommend fixing the issue a different way -- there's no reason the function has to work with vectors, function pointers, and ints. Why can't it work with linked lists, functors, and doubles? There's really no reason to constrain the types like this; just let them be whatever they are and see if the instantiation succeeds:
template <class TContainer, TFunction>
void map(TContainer & values, TFunction const & func) {
for (auto &value : values) {
value = func(value);
}
}
Expanded from my comment (when the question was closed), you can template away the function details using the functor template "pattern":
template <class T, typename Functor>
void map(std::vector<T>& values, Functor func) {
for (int &value : values) {
value = func(value);
}
}
See here for full example: https://godbolt.org/z/fdHvAP
I'm experiencing some problems which can be resumed by the following piece of code:
template <typename Key, typename Data, typename fct>
size_t wrapper(const std::pair<Key, Data> & p)
{
return fct(p.first);
}
int main(int argc, char *argv[])
{
size_t val =
wrapper<int, int, dft_hash_fct<int>>(std::pair<int,int>(5,9));
return 0;
}
I'm using clang compiler version 3.4 and this code does not compile with the following error
test-tmp.C:17:5: error: no matching function for call to 'wrapper'
wrapper<int, int, dft_hash_fct<int>>(std::pair<int,int>(5,9));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
test-tmp.C:9:8: note: candidate template ignored: invalid explicitly-specified argument
for template parameter 'fct'
The idea is to wrap a hash function (the template parameter fct) on std::pair for only taking the first field.
dft_hash_fct is another template defines as follows:
template <typename Key>
size_t dft_hash_fct(const Key & key)
{
return SuperFastHash(key);
}
This generic function works; it has been used in other contexts.
The purpose of all this is to reuse a hash based set (not map) as a map of keys to items of any type. The hash based ser receives a hash function in construction time.
Thanks for your comments (David, Andrey and Kazark)
Edited:
Well, I see, typename fct is a type, so I cannot handle as a pointer function; sorry for the trivia. Unfortunately, I believe that the approach of passing the function as parameter in the wrapper does not work, because the hash set expects a function pointer with the following signature:
size_t (*the_function)(const Key & key);
So, realizing this, thanks to your observations, I changed the code in question to:
template <typename Key, typename Data, size_t (*fct)(const Key & k)>
size_t wrapper(const std::pair<Key, Data> & p)
{
return (*fct)(p.first);
}
int main(int argc, char *argv[])
{
size_t val =
wrapper<int, int, dft_hash_fct<int>>(std::pair<int,int>(5,9));
return 0;
}
that compiles, links and runs. In addition, I put this line:
size_t (*fct)(const std::pair<int, int>&) =
wrapper<int, int, dft_hash_fct<int>>;
cout << (*fct)(std::pair<int, int>(4,6)) << endl;
And that compiles, links ans runs too. So, I can say that the compiler (and of course according to the language) can instantiate the function and handle a function pointer to it.
So, after that I tried to modify my original code, which is a derived class of HAshSet intended for managing pairs hashed by first field.
I declare some as:
template <typename Key, typename Data>
class HashMap : public HashSet<std::pair<Key, Data>>
{
...
HashMap(size_t (*function)(const Key & key))
: HashSet<Key, Data>(wrapper<Key, Data, function>)
{
}
..
};
But the compilation (with std=c++11) fails with the error
./tpl_dynSetHash.H:353:7: error: no matching constructor for initialization of
'HashSet<std::pair<unsigned long, long>>'
: HashSet<std::pair<Key,Data>(
^
testDynSetHash.C:178:8: note: in instantiation of member function
'HashMap<unsigned long, long>::HashMap' requested here
HMap table;
However, if I substitute the call to base constructor by
: HashSet<Key, Data>(wrapper<Key, Data, dft_hash_fct<Key>)
That compiles fine. Thus, I believe that the problem is with the parameter type declaration (but I do not know what is).
The standard idiom to pass functions is to pass them as function objects, e.g.
template <typename Key, typename Data, typename Fct>
size_t wrapper(const std::pair<Key, Data> & p, Fct fct)
{
return fct(p.first);
}
Then call the wrapper using:
int main(int argc, char *argv[])
{
// no explicit template arguments required
size_t val =
wrapper(std::pair<int,int>(5,9), &dft_hash_fct<int>);
return 0;
}
In your code, on the other hand:
template <typename Key, typename Data, typename fct>
size_t wrapper(const std::pair<Key, Data> & p)
{
return fct(p.first);
}
typename fct introduces an alias for a type. Inside this function, fct names a type; therefore fct(p.first) creates an object of type fct, and this object needs to be converted to a size_t in order to return it from wrapper. You can use this as well, but the type you had to use would have to look like this:
struct dft_hash_fct_t
{
size_t result;
dft_hash_fct_t(int p) : result(SuperFashHash(p)) {}
operator size_t() const { return result; }
};
Which is probably not what you intended.
The template declaration in
template <typename Key, typename Data, typename fct>
size_t wrapper(const std::pair<Key, Data> & p)
{
return fct(p.first);
}
declares template parameterfct as a type, but you are trying to pass a function pointer to it. You can make fct function pointer template parameter like this:
template <typename Key, typename Data, size_t(*fct)(const Key&)>
size_t wrapper(const std::pair<Key, Data> & p)
{
return fct(p.first);
}
However, the more idiomatic way is (as DyP says) to pass a function object so that the function works with function pointers as well as objects overloading operator():
template <typename Key, typename Data, typename Fct>
size_t wrapper(const std::pair<Key, Data> & p, Fct fct)
{
return fct(p.first);
}
Then when calling it you pass the function as a parameter
wrapper(std::pair<int,int>(5,9), dft_hash_fct<int>);
The code you wrote makes no sense within the context of your intent. Your fct template parameter is a type. That means that
return fct(p.first);
is a function-style cast, not an application of () operator (i.e. it is not a function call). In your code you are attempting to cast p.first to type fct and then attempting to return the result of that cast as size_t. Was that your intent? I doubt that it was. On top of that you are trying to pass a function pointer value dft_hash_fct<int> as a template argument for fct, i.e. you are passing a value where a type is expected. How did you expect it to work?
The description you provided seems to imply that you actually wanted to call a functor with type fct from inside wrapper instead of performing a cast. In order to do that you have to obtain the functor itself somehow. Remember again that fct is not a functor, its is just the type of the functor.
The typical approach would be to pass the functor from the outside, as function parameter
template <typename Key, typename Data, typename fct>
size_t wrapper(const std::pair<Key, Data> & p, fct f)
{
return f(p.first);
}
Now you can use your wrapper template with class-based functors, as well as with ordinary functions
size_t val = wrapper(std::pair<int,int>(5,9), dft_hash_fct<int>);
Note that dft_hash_fct<int> has to be supplied as function argument, not as template argument.
There's no need to explicitly specify template arguments, since they will be deduced by the compiler.
This used to work some weeks ago:
template <typename T, T t>
T tfunc()
{
return t + 10;
}
template <typename T>
constexpr T func(T t)
{
return tfunc<T, t>();
}
int main()
{
std::cout << func(10) << std::endl;
return 0;
}
But now g++ -std=c++0x says:
main.cpp: In function ‘constexpr T func(T) [with T = int]’:
main.cpp:29:25: instantiated from here
main.cpp:24:24: error: no matching function for call to ‘tfunc()’
main.cpp:24:24: note: candidate is:
main.cpp:16:14: note: template<class T, T t> T tfunc()
main.cpp:25:1: warning: control reaches end of non-void function [-Wreturn-type]
clang++ -std=c++11 says that template's parameters of tfunc<T, t>() are ignored because invalid.
Is that a bug, or a fix ?
PS:
g++ --version => g++ (GCC) 4.6.2 20120120 (prerelease)
clang++ --version => clang version 3.0 (tags/RELEASE_30/final) (3.0.1)
The parameter t is not a constant expression. Hence the error. It should be also noted that it cannot be a constant expression.
You can pass the constant expression as argument, but inside the function, the object (the parameter) which holds the value, is not a constant expression.
Since t is not a constant expression, it cannot be used as template argument:
return tfunc<T, t>(); //the second argument must be a constant expression
Maybe, you want something like this:
template <typename T, T t>
T tfunc()
{
return t + 10;
}
template <typename T, T t> //<---- t became template argument!
constexpr T func()
{
return tfunc<T, t>();
}
#define FUNC(a) func<decltype(a),a>()
int main()
{
std::cout << FUNC(10) << std::endl;
}
Now it should work : online demo
I get the feeling that constexpr must also be valid in a 'runtime' context, not just at compile-time. Marking a function as constexpr encourages the compiler to try to evaluate it at compile-time, but the function must still have a valid run-time implementation.
In practice, this means that the compiler doesn't know how to implement this function at runtime:
template <typename T>
constexpr T func(T t)
{
return tfunc<T, t>();
}
A workaround is to change the constructor such that it takes its t parameter as a normal parameter, not as a template parameter, and mark the constructor as constexpr:
template <typename T>
constexpr T tfunc(T t)
{
return t + 10;
}
template <typename T>
constexpr T func(T t)
{
return tfunc<T>(t);
}
There are three levels of 'constant-expression-ness':
template int parameter, or (non-VLA) array size // Something that must be a constant-expression
constexpr // Something that may be a constant-expression
non-constant-expression
You can't really convert items that are low in that list into something that is high in that list, but obviously the other route it possible.
For example, a call to this function
constexpr int foo(int x) { return x+1; }
isn't necessarily a constant-expression.
// g++-4.6 used in these few lines. ideone doesn't like this code. I don't know why
int array[foo(3)]; // this is OK
int c = getchar();
int array[foo(c)]; // this will not compile (without VLAs)
So the return value from a constexpr function is a constant expression only if all the parameters, and the implementation of the function, can be completed at executed at compile-time.
Recap the question: You have two functions which take a parameter of type T. One takes its parameter as a template parameter, and the other as a 'normal' parameter.
I'm going to call the two functions funcT and funcN instead of tfunc and func.
You wish to be able to call funcT from funcN. Marking the latter as a constexpr doesn't help.
Any function marked as constexpr must be compilable as if the constexpr wasn't there. constexpr functions are a little schizophrenic. They only graduate to full constant-expressions in certain circumstances.
It would not be possible to implement funcN to run at runtime in a simple way, as it would need to be able to work for all possible values of t. This would require the compiler to instantiate many instances of tfunc, one for each value of t. But you can work around this if you're willing to live with a small subset of T. There is a template-recursion limit of 1024 in g++, so you can easily handle 1024 values of T with this code:
#include<iostream>
#include<functional>
#include<array>
using namespace std;
template <typename T, T t>
constexpr T funcT() {
return t + 10;
}
template<typename T, T u>
constexpr T worker (T t) {
return t==0 ? funcT<T,u>() : worker<T, u+1>(t-1);
}
template<>
constexpr int worker<int,1000> (int ) {
return -1;
}
template <typename T>
constexpr T funcN(T t)
{
return t<1000 ? worker<T,0>(t) : -1;
}
int main()
{
std::cout << funcN(10) << std::endl;
array<int, funcN(10)> a; // to verify that funcN(10) returns a constant-expression
return 0;
}
It uses a function worker which will recursively convert the 'normal' parameter t into a template parameter u, which it then uses to instantiate and execute tfunc<T,u>.
The crucial line is return funcT<T,u>() : worker<T, u+1>(t-1);
This has limitations. If you want to use long, or other integral types, you'll have to add another specialization. Obviously, this code only works for t between 0 and 1000 - the exact upper limit is probably compiler-dependent. Another option might be to use a binary search of sorts, with a different worker function for each power of 2:
template<typename T, T u>
constexpr T worker4096 (T t) {
return t>=4096 ? worker2048<T, u+4096>(t-4096) : worker2048<T, u>(t);
}
I think this will work around the template-recursion-limit, but it will still require a very large number of instantiations and would make compilation very slow, if it works at all.
Looks like it should give an error - it has no way of knowing that you passed in a constant value as t to func.
More generally, you can't use runtime values as template arguments. Templates are inherently a compile-time construct.
Is it a bug in Visual C++ 2010 or right behaviour?
template<class T>
T f(T const &r)
{
return r;
}
template<class T>
T f(T &&r)
{
static_assert(false, "no way"); //< line # 10
return r;
}
int main()
{
int y = 4;
f(y); //< line # 17
}
I thought, the function f(T &&) should never be called but it's called with T = int &. The output:
main.cpp(10): error C2338: no way
main.cpp(17) : see reference to function template instantiation 'T f(T)' being compiled
with
[
T=int &
]
Update 1 Do you know any C++x0 compiler as a reference? I've tried comeau online test-drive but could not compile r-value reference.
Update 2 Workaround (using SFINAE):
#include <boost/utility/enable_if.hpp>
#include <boost/type_traits/is_reference.hpp>
template<class T>
T f(T &r)
{
return r;
}
template<class T>
typename ::boost::disable_if< ::boost::is_reference<T>, T>::type f(T &&r)
{
static_assert(false, "no way");
return r;
}
int main()
{
int y = 4;
f(y);
// f(5); // generates "no way" error, as expected.
}
Update 3 Some of compilers trigger on static_assert(false, "no way") even if no function template instantiation. Workaround (thanks to #Johannes Schaub - litb)
template<class T> struct false_ { static bool const value = false; };
...
static_assert(false_<T>::value, "no way");
or
static_assert(sizeof(T) == sizeof(T), "no way");
As I understand it (and I may not be completely right; the specification is a bit complicated), the template type deduction rules conspire against you.
The compiler first attempts to substitute all templates (it's not choosing at this point yet—just looking for options) and gets:
T const &r matches int lvalue with T = int, creating f(int const &)
T &&r matches int lvalue with T = int& and int & && reduces to int&, creating f(int &) (there are rules saying this in the spec).
Now it comes to selecting correct overload and the later is better match, because the first differs in cv-qualification and the later does not. That's also the reason why when you remove the const, you get ambiguous overload error—the overloads end up being exactly the same.
Ad Update1: gcc supports many of the C++0x features. You can get native windows build from mingw or use cygwin.
Ad Update2: If you really need separate overloads for rvalue and lvalue, that seems to be the only option. But most templates do the right thing with just any kind of reference, perhaps using std::forward to ensure proper resolution of functions they call depending on whether they got rvalue or lvalue).
Your fix doesn't solve the problem with static_assert firing though. The static_assert(false, ...) will still trigger for compilers that parse templates at definition time (most do).
They will see that any function template instantiation will be ill-formed, and the Standard allows them to issue an error for the template itself then, and most will do so.
For making this work you need to make the expression dependent so that the compiler doesn't know when it parses the template that it will always evaluate to false. For example
template<class> struct false_ { static bool const value = false; };
template<class T>
T f(T &&r)
{
static_assert(false_<T>::value, "no way"); //< line # 10
return r;
}
This is a followup question to my previous question.
#include <functional>
int foo(void) {return 2;}
class bar {
public:
int operator() (void) {return 3;};
int something(int a) {return a;};
};
template <class C> auto func(C&& c) -> decltype(c()) { return c(); }
template <class C> int doit(C&& c) { return c();}
template <class C> void func_wrapper(C&& c) { func( std::bind(doit<C>, std::forward<C>(c)) ); }
int main(int argc, char* argv[])
{
// call with a function pointer
func(foo);
func_wrapper(foo); // error
// call with a member function
bar b;
func(b);
func_wrapper(b);
// call with a bind expression
func(std::bind(&bar::something, b, 42));
func_wrapper(std::bind(&bar::something, b, 42)); // error
// call with a lambda expression
func( [](void)->int {return 42;} );
func_wrapper( [](void)->int {return 42;} );
return 0;
}
I'm getting a compile errors deep in the C++ headers:
functional:1137: error: invalid initialization of reference of type ‘int (&)()’ from expression of type ‘int (*)()’
functional:1137: error: conversion from ‘int’ to non-scalar type ‘std::_Bind<std::_Mem_fn<int (bar::*)(int)>(bar, int)>’ requested
func_wrapper(foo) is supposed to execute func(doit(foo)). In the real code it packages the function for a thread to execute. func would the function executed by the other thread, doit sits in between to check for unhandled exceptions and to clean up. But the additional bind in func_wrapper messes things up...
At the beginning, please let me introduce 2 key points:
a: When using nested std::bind, the inner std::bind is evaluated first, and the return value will be substituted in its place while the outer std::bind is evaluated. That means std::bind(f, std::bind(g, _1))(x) executes as same as f(g(x)) does. The inner std::bind is supposed to be wrapped by std::ref if the outer std::bind wants a functor rather than a return value.
b: The r-value reference cannot be correctly forwarded to the function by using std::bind. And the reason has already been illustrated in detail.
So, let's look at the question. The most importance function here might be func_wrapper which is intended to perform 3 purposes:
Perfect forwarding a functor to doit function template at first,
then using std::bind to make doit as a closure,
and letting func function template execute the functor returned by std::bind at last.
According to point b, purpose 1 cannot be fulfilled. So, let's forget perfect forwarding and doit function template has to accept a l-value reference parameter.
According to point a, purpose 2 will be performed by using std::ref.
As a result, the final version might be:
#include <functional>
int foo(void) {return 2;}
class bar {
public:
int operator() (void) {return 3;};
int something(int a) {return a;};
};
template <class C> auto func(C&& c) -> decltype(c()) { return c(); }
template <class C> int doit(C&/*&*/ c) // r-value reference can't be forwarded via std::bind
{
return c();
}
template <class C> void func_wrapper(C&& c)
{
func(std::bind(doit<C>,
/* std::forward<C>(c) */ // forget pefect forwarding while using std::bind
std::ref(c)) // try to pass the functor itsself instead of its return value
);
}
int main(int argc, char* argv[])
{
// call with a function pointer
func(foo);
func_wrapper(foo); // error disappears
// call with a member function
bar b;
func(b);
func_wrapper(b);
// call with a bind expression
func(std::bind(&bar::something, b, 42));
func_wrapper(std::bind(&bar::something, b, 42)); // error disappears
// call with a lambda expression
func( [](void)->int {return 42;} );
func_wrapper( [](void)->int {return 42;} );
return 0;
}
But, if you really want to achieve purpose 1 and 2, how? Try this:
#include <functional>
#include <iostream>
void foo()
{
}
struct bar {
void operator()() {}
void dosomething() {}
};
static bar b;
template <typename Executor>
void run(Executor&& e)
{
std::cout << "r-value reference forwarded\n";
e();
}
template <typename Executor>
void run(Executor& e)
{
std::cout << "l-value reference forwarded\n";
e();
}
template <typename Executor>
auto func(Executor&& e) -> decltype(e())
{
return e();
}
template <bool b>
struct dispatcher_traits {
enum { value = b };
};
template <typename Executor, bool is_lvalue_reference>
class dispatcher {
private:
static void dispatch(Executor& e, dispatcher_traits<true>)
{
run(e);
}
static void dispatch(Executor& e, dispatcher_traits<false>)
{
run(std::ref(e));
}
public:
static void forward(Executor& e)
{
dispatch(e, dispatcher_traits<is_lvalue_reference>());
}
};
template <typename Executor>
void func_wrapper(Executor&& e)
{
typedef dispatcher<Executor,
std::is_lvalue_reference<Executor>::value>
dispatcher_type;
func(std::bind(&dispatcher_type::forward, std::ref(e)));
}
int main()
{
func_wrapper(foo); // l-value
func_wrapper(b); // l-value
func_wrapper(bar()); // r-value
func_wrapper(std::bind(&bar::dosomething, &b)); // r-value
func_wrapper([](){}); // r-value
}
Let me explain some points:
To reduce lots of return statements, changing functor signature from int() to void().
The 2 run() function templates are used to check whether the original functor parameter is perfect forwarded or not.
dispatcher_traits is going to map bool constant to type.
You'd better name dispatcher::forward to differ from dispatcher::dispatch or you have to invoke std::bind template with dispatcher::forward's signature.
Looking at this the second time now, and I think I have a plausable explanation for the first error you are seeing.
In this case, it's more helpful to look at the complete error and the template instantiations that lead up to it. The error printed by my compiler (GCC 4.4), for example, ends with the following lines:
test.cpp:12: instantiated from ‘decltype (c()) func(C&&) [with C = std::_Bind<int (*(int (*)()))(int (&)())>]’
test.cpp:16: instantiated from ‘void func_wrapper(C&&) [with C = int (&)()]’
test.cpp:22: instantiated from here
/usr/include/c++/4.4/tr1_impl/functional:1137: error: invalid initialization of reference of type ‘int (&)()’ from expression of type ‘int (*)()’
Now looking at this bottom-up, the actual error message seems correct; the types the compiler has deduced are incompatible.
The first template instantiation, at func_wrapper, clearly shows what type the compiler has deduced from the actual parameter foo in func_wrapper(foo). I personally expected this to be a function pointer, but it is in fact a function reference.
The second template instantiation is hardly readable. But messing around with std::bind a bit, I learned that the format of the textual representation GCC prints for a bind functor is roughly:
std::_Bind<RETURN-TYPE (*(BOUND-VALUE-TYPES))(TARGET-PARAMETER-TYPES)>
So tearing it apart:
std::_Bind<int (*(int (*)()))(int (&)())>
// Return type: int
// Bound value types: int (*)()
// Target parameter types: int (&)()
This is where the incompatible types start. Apparently, even though c in func_wrapper is a function reference, it turns into a function pointer once passed to std::bind, resulting in the type incompatibility. For what it's worth, std::forward doesn't matter at all in this case.
My reasoning here is that std::bind only seems to care about values, and not references. In C/C++, there's no such thing as a function value; there's only references and pointers. So when the function reference is dereferenced, the compiler can only meaningfully give you a function pointer.
The only control you have over this is your template parameters. You will have to tell the compiler that you're dealing with a function pointer from the start to make this work. It's probably what you had in mind anyways. To do that, explicitly specify the type you want for the template parameter C:
func_wrapper<int (*)()>(foo);
Or the more brief solution, explicitly take the function's address:
func_wrapper(&foo); // with C = int (*)()
I'll get back to you if I ever figure out the second error. :)