Suppose I have a callable type like so:
struct mutable_callable
{
int my_mutable = 0;
int operator()() { // Not const
return my_mutable++;
}
};
Note that mutable_callable has a non-const operator() that modifies a member variable.....
Now suppose I create a std::function out of my type:
std::function<int()> foo = mutable_callable{};
Now I can do this:
void invoke(std::function<int()> const& z)
{
z();
}
int main()
{
invoke(foo); // foo changed.....oops
}
Now as far as I can tell std::functions operator() is const as per:
https://en.cppreference.com/w/cpp/utility/functional/function/operator()
So my gut feeling is that you shouldn't be able to do this.....
But then looking at:
https://en.cppreference.com/w/cpp/utility/functional/function/function
This doesn't seem to put any constraints on whether or not the callable type has a constant operator()......
So my question is this: I am correct in assuming that std::function<int()> const& is essentially the same thing as std::function<int()>& that is there is no actually difference between the behavior of the two......and if that is the case why is it not const correct?
This boils down to the same as struct A { int* x; };, where in a const A a; you can modify the value of *(a.x) (but not where it points to). There is a level of indirection in std::function (from the type erasure) through which const is not propagated.
And no, std::function<int()> const& f is not pointless. In a std::function<int()>& f you would be able to assign a different functor to f, which you cannot do in the const case.
Related
What would be a valid use case for a signature like this?:
T&& foo();
Or is the rvalue ref only intended for use as argument?
How would one use a function like this?
T&& t = foo(); // is this a thing? And when would t get destructed?
For a free function it doesn't make much sense to return a rvalue reference. If it is a non-static local object then you never want to return a reference or pointer to it because it will be destroyed after the function returns. It can possibly make sense to return a rvalue reference to an object that you passed to the function though. It really depends on the use case for if it makes sense or not.
One thing that can greatly benefit from returning an rvalue reference is a member function of a temporary object. Lets say you have
class foo
{
std::vector<int> bar;
public:
foo(int n) : bar(n) {}
std::vector<int>& get_vec() { return bar; }
};
If you do
auto vec = foo(10).get_vec();
you have to copy because get_vec returns an lvalue. If you instead use
class foo
{
std::vector<int> bar;
public:
foo(int n) : bar(n) {}
std::vector<int>& get_vec() & { return bar; }
std::vector<int>&& get_vec() && { return std::move(bar); }
};
Then vec would be able to move the vector returned by get_vec and you save yourself an expensive copy operation.
T&& t = foo(); // is this a thing? And when would t get destructed?
An rvalue reference is really similar to a lvalue reference. Think about your example like it was normal references:
T& foo();
T& t = foo(); // when is t destroyed?
The answer is that t is still valid to use as long as the object is refers to lives.
The same answer still applies to you rvalue reference example.
But... does it make sense to return an rvalue reference?
Sometimes, yes. But very rarely.
consider this:
std::vector<int> v = ...;
// type is std::tuple<std::vector<int>&&>
auto parameters = std::forward_as_tuple(std::move(v));
// fwd is a rvalue reference since std::get returns one.
// fwd is valid as long as v is.
decltype(auto) fwd = std::get<0>(std::move(parameters));
// useful for calling function in generic context without copying
consume(std::get<0>(std::move(parameters)));
So yes there are example. Here, another interesting one:
struct wrapper {
auto operator*() & -> Heavy& {
return heavy;
}
auto operator*() && -> Heavy&& {
return std::move(heavy);
}
private:
Heavy instance;
};
// by value
void use_heavy(Heavy);
// since the wrapper is a temporary, the
// Heavy contained will be a temporary too.
use_heavy(*make_wrapper());
I think a use case would be to explicitly give permission to "empty" some non-local variable. Perhaps something like this:
class Logger
{
public:
void log(const char* msg){
logs.append(msg);
}
std::vector<std::string>&& dumpLogs(){
return std::move(logs);
}
private:
std::vector<std::string> logs;
};
But I admit I made this up now, I never actually used it and it also can be done like this:
std::vector<std::string> dumpLogs(){
auto dumped_logs = logs;
return dumped_logs;
}
This question already has answers here:
How do I remove code duplication between similar const and non-const member functions?
(21 answers)
C++ template to cover const and non-const method
(7 answers)
Closed 5 years ago.
Is there any advantage using one over the other:
class Foo
{
public:
const int& get() const
{
// stuff here
return myInt;
}
int& get()
{
return const_cast<int&>(static_cast<const Foo*>(this)->get());
}
};
Or
class Foo
{
public:
int& get()
{
// stuff here
return myInt;
}
const int& get() const
{
return const_cast<Foo*>(this)->get();
}
};
I only used the first one, but I just saw the second one used somewhere, so I am wondering.
The comment // stuff here could be a non-trivial check like retrieving the index of a table in order to return a ref on a member of the table (for example: myInt = myTable[myComputedIndex];) so I cannot just make it public. Thus table and any member are not const.
If you have to make a function that is const-agnostic, and avoids duplication, one neat way to do it is delegating implementation to a template, for example
class Foo {
private:
int my_int;
template <typename ThisPtr>
static auto& get(ThisPtr this_ptr) {
return this_ptr->my_int;
}
public:
int& get() {
return get(this);
}
const int& get() const {
return get(this);
}
};
This way you are free from the fear associated with using const_cast, mutable and other stuff that goes into trying to reduce code duplication in cases like this. If you get something wrong, the compiler will let you know.
Ignoring the issue of whether you really need a getter, the best solution when duplicating functionality in both a const and non-const method is to have the non-const method call the const method and cast away the const-ness of the result (i.e. the first of the two alternatives you present in the question).
The reason is simple: if you do it the other way around (with the logic in the non-const method), you could accidentally end up modifying a const object, and the compiler won't catch it at compile time (because the method is not declared const) - this will have undefined behaviour.
Of course this is only a problem if the "getter" is not actually a getter (i.e. if it is doing something more complicated than just returning a reference to a private field).
Also, if you are not constrained to C++11, the template-based solution presented by Curious in their answer is another way of avoiding this problem.
Is there any advantage using one over the other: ...
No, both are bad because they violate the data encapsulation principle.
In your example you should rather make myInt a public member.
There's no advantage to have getters for such case at all.
If you really want (need) getter and setter functions these should look like this:
class Foo
{
private:
mutable int myInt_;
// ^^^^^^^ Allows lazy initialization from within the const getter,
// simply omit that if you dont need it.
public:
void myInt(int value)
{
// Do other stuff ...
myInt = value;
// Do more stuff ...
}
const int& myInt() const
{
// Do other stuff ...
return myInt_;
}
}
You don't say where myInt comes from, the best answer depends on that.
There are 2+1 possible scenarios:
1) The most common case is that myInt comes from a pointer internal to the class.
Assuming that, this is the best solution which avoids both code duplication and casting.
class Foo{
int* myIntP;
...
int& get_impl() const{
... lots of code
return *myIntP; // even if Foo instance is const, *myInt is not
}
public:
int& get(){return get_impl();}
const int& get() const{return get_impl();}
};
This case above applies to pointer array, and (most) smart pointers.
2) The other common case is that myInt is a reference or a value member, then the previous solution doesn't work.
But it is also the case where a getter is not needed at all.
Don't use a getter in that case.
class Foo{
public:
int myInt; // or int& myInt;
};
done! :)
3) There is a third scenario, pointed by #Aconcagua, that is the case of an internal fixed array. In that case it is a toss-up, it really depends what you are doing, if finding the index is really the problem, then that can be factored away. It is not clear however what is the application:
class Foo{
int myInts[32];
...
int complicated_index() const{...long code...}
public:
int& get(){return myInts[complicated_index()];}
const int& get() const{return myInts[complicated_index()];}
};
My point is, understand the problem and don´t over engineer. const_cast or templates are not needed to solve this problem.
complete working code below:
class Foo{
int* myIntP;
int& get_impl() const{
return *myIntP; // even if Foo instance is const, *myInt is not
}
public:
int& get(){return get_impl();}
const int& get() const{return get_impl();}
Foo() : myIntP(new int(0)){}
~Foo(){delete myIntP;}
};
#include<cassert>
int main(){
Foo f1;
f1.get() = 5;
assert( f1.get() == 5 );
Foo const f2;
// f2.get() = 5; // compile error
assert( f2.get() == 0 );
return 0;
}
As you intend access to more complex internal structures (as clarified via your edit; such as providing an operator[](size_t index) for internal arrays as std::vector does), then you will have to make sure that you do not invoke undefined behaviour by modifying a potentially const object.
The risk of doing so is higher in the second approach:
int& get()
{
// stuff here: if you modify the object, the compiler won't warn you!
// but you will modify a const object, if the other getter is called on one!!!
return myInt;
}
In the first variant, you are safe from (unless you do const_cast here, too, which now would really be bad...), which is the advantage of this approach:
const int& get() const
{
// stuff here: you cannot modify by accident...
// if you try, the compiler will complain about
return myInt;
}
If you actually need to modify the object in the non-const getter, you cannot have a common implementation anyway...
Modifying a const object through a non-const access path [...] results in undefined behavior.
(Source: http://en.cppreference.com/w/cpp/language/const_cast)
This means that the first version can lead to undefined behavior if myInt is actually a const member of Foo:
class Foo
{
int const myInt;
public:
const int& get() const
{
return myInt;
}
int& get()
{
return const_cast<int&>(static_cast<const Foo*>(this)->get());
}
};
int main()
{
Foo f;
f.get() = 10; // this compiles, but it is undefined behavior
}
The second version would not compile, because the non-const version of get would be ill-formed:
class Foo
{
int const myInt;
public:
int& get()
{
return myInt;
// this will not compile, you cannot return a const member
// from a non-const member function
}
const int& get() const
{
return const_cast<Foo*>(this)->get();
}
};
int main()
{
Foo f;
f.get() = 10; // get() is ill-formed, so this does not compile
}
This version is actually recommended by Scott Meyers in Effective C++ under Avoid Duplication in const and Non-const Member Function.
Imagine you have a number of overloaded methods that (before C++11) looked like this:
class MyClass {
public:
void f(const MyBigType& a, int id);
void f(const MyBigType& a, string name);
void f(const MyBigType& a, int b, int c, int d);
// ...
};
This function makes a copy of a (MyBigType), so I want to add an optimization by providing a version of f that moves a instead of copying it.
My problem is that now the number of f overloads will duplicate:
class MyClass {
public:
void f(const MyBigType& a, int id);
void f(const MyBigType& a, string name);
void f(const MyBigType& a, int b, int c, int d);
// ...
void f(MyBigType&& a, int id);
void f(MyBigType&& a, string name);
void f(MyBigType&& a, int b, int c, int d);
// ...
};
If I had more parameters that could be moved, it would be unpractical to provide all the overloads.
Has anyone dealt with this issue? Is there a good solution/pattern to solve this problem?
Thanks!
Herb Sutter talks about something similar in a cppcon talk
This can be done but probably shouldn't. You can get the effect out using universal references and templates, but you want to constrain the type to MyBigType and things that are implicitly convertible to MyBigType. With some tmp tricks, you can do this:
class MyClass {
public:
template <typename T>
typename std::enable_if<std::is_convertible<T, MyBigType>::value, void>::type
f(T&& a, int id);
};
The only template parameter will match against the actual type of the parameter, the enable_if return type disallows incompatible types. I'll take it apart piece by piece
std::is_convertible<T, MyBigType>::value
This compile time expression will evaluate to true if T can be converted implicitly to a MyBigType. For example, if MyBigType were a std::string and T were a char* the expression would be true, but if T were an int it would be false.
typename std::enable_if<..., void>::type // where the ... is the above
this expression will result in void in the case that the is_convertible expression is true. When it's false, the expression will be malformed, so the template will be thrown out.
Inside the body of the function you'll need to use perfect forwarding, if you are planning on copy assigning or move assigning, the body would be something like
{
this->a_ = std::forward<T>(a);
}
Here's a coliru live example with a using MyBigType = std::string. As Herb says, this function can't be virtual and must be implemented in the header. The error messages you get from calling with a wrong type will be pretty rough compared to the non-templated overloads.
Thanks to Barry's comment for this suggestion, to reduce repetition, it's probably a good idea to create a template alias for the SFINAE mechanism. If you declare in your class
template <typename T>
using EnableIfIsMyBigType = typename std::enable_if<std::is_convertible<T, MyBigType>::value, void>::type;
then you could reduce the declarations to
template <typename T>
EnableIfIsMyBigType<T>
f(T&& a, int id);
However, this assumes all of your overloads have a void return type. If the return type differs you could use a two-argument alias instead
template <typename T, typename R>
using EnableIfIsMyBigType = typename std::enable_if<std::is_convertible<T, MyBigType>::value,R>::type;
Then declare with the return type specified
template <typename T>
EnableIfIsMyBigType<T, void> // void is the return type
f(T&& a, int id);
The slightly slower option is to take the argument by value. If you do
class MyClass {
public:
void f(MyBigType a, int id) {
this->a_ = std::move(a); // move assignment
}
};
In the case where f is passed an lvalue, it will copy construct a from its argument, then move assign it into this->a_. In the case that f is passed an rvalue, it will move construct a from the argument and then move assign. A live example of this behavior is here. Note that I use -fno-elide-constructors, without that flag, the rvalue cases elides the move construction and only the move assignment takes place.
If the object is expensive to move (std::array for example) this approach will be noticeably slower than the super-optimized first version. Also, consider watching this part of Herb's talk that Chris Drew links to in the comments to understand when it could be slower than using references. If you have a copy of Effective Modern C++ by Scott Meyers, he discusses the ups and downs in item 41.
You may do something like the following.
class MyClass {
public:
void f(MyBigType a, int id) { this->a = std::move(a); /*...*/ }
void f(MyBigType a, string name);
void f(MyBigType a, int b, int c, int d);
// ...
};
You just have an extra move (which may be optimized).
My first thought is that you should change the parameters to pass by value. This covers the existing need to copy, except the copy happens at the call point rather than explicitly in the function. It also allows the parameters to be created by move construction in a move-able context (either unnamed temporaries or by using std::move).
Why you would do that
These extra overloads only make sense, if modifying the function paramers in the implementation of the function really gives you a signigicant performance gain (or some kind of guarantee). This is hardly ever the case except for the case of constructors or assignment operators. Therefore, I would advise you to rethink, whether putting these overloads there is really necessary.
If the implementations are almost identical...
From my experience this modification is simply passing the parameter to another function wrapped in std::move() and the rest of the function is identical to the const & version. In that case you might turn your function into a template of this kind:
template <typename T> void f(T && a, int id);
Then in the function implementation you just replace the std::move(a) operation with std::forward<T>(a) and it should work. You can constrain the parameter type T with std::enable_if, if you like.
In the const ref case: Don't create a temporary, just to to modify it
If in the case of constant references you create a copy of your parameter and then continue the same way the move version works, then you may as well just pass the parameter by value and use the same implementation you used for the move version.
void f( MyBigData a, int id );
This will usually give you the same performance in both cases and you only need one overload and implementation. Lots of plusses!
Significantly different implementations
In case the two implementations differ significantly, there is no generic solution as far as I know. And I believe there can be none. This is also the only case, where doing this really makes sense, if profiling the performance shows you adequate improvements.
You might introduce a mutable object:
#include <memory>
#include <type_traits>
// Mutable
// =======
template <typename T>
class Mutable
{
public:
Mutable(const T& value) : m_ptr(new(m_storage) T(value)) {}
Mutable(T& value) : m_ptr(&value) {}
Mutable(T&& value) : m_ptr(new(m_storage) T(std::move(value))) {}
~Mutable() {
auto storage = reinterpret_cast<T*>(m_storage);
if(m_ptr == storage)
m_ptr->~T();
}
Mutable(const Mutable&) = delete;
Mutable& operator = (const Mutable&) = delete;
const T* operator -> () const { return m_ptr; }
T* operator -> () { return m_ptr; }
const T& operator * () const { return *m_ptr; }
T& operator * () { return *m_ptr; }
private:
T* m_ptr;
char m_storage[sizeof(T)];
};
// Usage
// =====
#include <iostream>
struct X
{
int value = 0;
X() { std::cout << "default\n"; }
X(const X&) { std::cout << "copy\n"; }
X(X&&) { std::cout << "move\n"; }
X& operator = (const X&) { std::cout << "assign copy\n"; return *this; }
X& operator = (X&&) { std::cout << "assign move\n"; return *this; }
~X() { std::cout << "destruct " << value << "\n"; }
};
X make_x() { return X(); }
void fn(Mutable<X>&& x) {
x->value = 1;
}
int main()
{
const X x0;
std::cout << "0:\n";
fn(x0);
std::cout << "1:\n";
X x1;
fn(x1);
std::cout << "2:\n";
fn(make_x());
std::cout << "End\n";
}
This is the critical part of the question:
This function makes a copy of a (MyBigType),
Unfortunately, it is a little ambiguous. We would like to know what is the ultimate target of the data in the parameter. Is it:
1) to be assigned to an object that existing before f was called?
2) or instead, stored in a local variable:
i.e:
void f(??? a, int id) {
this->x = ??? a ???;
...
}
or
void f(??? a, int id) {
MyBigType a_copy = ??? a ???;
...
}
Sometimes, the first version (the assignment) can be done without any copies or moves. If this->x is already long string, and if a is short, then it can efficiently reuse the existing capacity. No copy-construction, and no moves. In short, sometimes assignment can be faster because we can skip the copy contruction.
Anyway, here goes:
template<typename T>
void f(T&& a, int id) {
this->x = std::forward<T>(a); // is assigning
MyBigType local = std::forward<T>(a); // if move/copy constructing
}
If the move version will provide any optimization then the implementation of the move overloaded function and the copy one must be really different. I don't see a way to get around this without providing implementations for both.
Lets say I have a class C, and a function make_c(x) which creates instances of C.
C stores x by reference.
How can I write make_c(x) to give a compile error when x is an unnamed temporary (that would of course destruct at the end of line, leaving a dangling reference) but accept named temporaries and other values?
I believe this should have the semantics you're looking for:
template<typename X>
C make_c(X&& x)
{
static_assert(
!std::is_rvalue_reference<decltype(std::forward<X>(x))>::value,
"x must not be a temporary"
);
return C(std::forward<X>(x));
}
Caveat: this won't work as-is with VC++ 2010 due to deficiencies in its implementation of decltype (you'd need to wrap decltype in std::identity<>).
I don't think it's possible within the language, because you'd need to check the flow control through arbitrary functions.
struct Foo{
};
Foo const & sanitize(Foo const & f){ return f;}
void checkThisFunction(Foo const & f){
//we'd like to ensure at compile time that f is not a temporary
}
int main(){
Foo f;
checkThisFunction(sanitize(f));
checkThisFunction(sanitize(Foo()));
return 0;
}
Unless I'm completely misunderstanding rvalue references, this sort of thing should be doable with simple overloading.
void foo(int&&) = delete;
void foo(const int&) { }
int main()
{
int a;
foo(a);
foo(42); //error, prefers binding to the deleted overload
}
Please consider the following code.
struct foo
{
};
template<typename T>
class test
{
public:
test() {}
const T& value() const
{
return f;
}
private:
T f;
};
int main()
{
const test<foo*> t;
foo* f = t.value();
return 0;
}
t is a const variable and value() is a constant member-function which returns const T&. AFAIK, a const type is not assignable to a non-const type. But how foo* f = t.value(); compiles well. How this is happening and how can I ensure value() can be only assigned to const foo*?
Edit
I found that, this is happening on when templates are used. Following code works as expected.
class test
{
public:
test() {}
const foo* value() const { return f; }
private:
foo* f;
};
int main()
{
const test t;
foo* f = t.value(); // error here
return 0;
}
Why the problem is happening when templates are used?
Because you have two levels of indirection - in your main function, that call to value returns a reference to a const pointer to a non-const foo.
This can safely be copied into non-const pointer to a non-const foo.
If you'd instantiated test with const foo *, it would be a different story.
const test<const foo*> t;
foo* f = t.value(); // error
const foo* f = t.value(); // fine
return 0;
Update
From the comment:
value() returns const T& which can
only be assigned to another const
type. But in this case, compiler is
safely allowing the conversion.
Const data can only be read. It cannot be written ("mutated"). But copying some data is a way of reading it, so it's okay. For example:
const int c = 5;
int n = c;
Here, I had some const data in c, and I copied the data into a non-const variable n. That's fine, it's just reading the data. The value in c has not been modified.
Now, suppose your foo had some data in it:
struct foo { int n; };
If I have a non-const pointer to one of those, I can modify the n value through the pointer. You asked your test template to store a pointer to a non-const foo, and then made a const instance of test. Only the pointer address is constant, therefore. No one can change the address stored in the pointer inside test, so it cannot be made to point to another object. However, the object it points to can have its contents modified.
Update 2:
When you made your non-template version of the example, you made a mistake. To get it right, you need to substitute foo * into each place where there's a T.
const T& value() const
Notice that you have a reference to a const T there. So the return value will be a reference to something const: a foo *. It's only the pointer address that can't be modified. The object it points to can have its contents modified.
In your second example, you got rid of the reference part, which changes the meaning and makes the const modifier apply to the object that the pointer points to, instead of applying to the pointer itself.
Use the following template specialization:
template<typename T>
class test<T*>
{
public:
test() {}
const T* value() const
{
return f;
}
private:
T* f;
};
After including this, g++ says:
d.cpp: In function ‘int main()’:
d.cpp:41: error: invalid conversion from ‘const foo*’ to ‘foo*’
There's nothing wrong in your code, having a const reference to a pointer only means that you can't modify the pointer, but the pointed-to object remains perfectly mutable. If inside your main function you try to change the address pointed to by the f member of t you'll see that you can't: encapsulation is perfectly preserved.
This is the same principle that makes the following code valid:
void foo(std::vector<int *> const & v)
{
*v[0] = 0; // op. [] returns const & to int *
}
People new to C++ are usually surprised by this behavior, because for them a const vector should not allow the modification of its elements. And in fact it doesn't, because the pointer stored in the vector does not change (it keeps pointing to the same address). It's the pointed-to object which is modified, but the vector does not care about that.
The only solution is to do as Amit says and provide a specialization of your class for T*.