The void(), the comma operator (operator,) and the impossible (?) overloading - c++

Consider the following struct:
struct S {};
In C++14, the definition below is valid:
constexpr auto f() { return S{}, 'c'; }
As well as the following one:
constexpr auto f() { return S{}, void(); }
Now, consider the following, working snippet that involves the first of the two definitions:
#include<type_traits>
struct S {};
constexpr int operator,(S, char) { return 42; }
constexpr auto f() { return S{}, 'c'; }
int main() {
constexpr int i{f()};
static_assert(i == 42, "!");
static_assert(std::is_same<decltype(f()), int>::value, "!");
}
Speaking not so technically, the overload of the comma operator intercepts the couple S{}, 'c' and returns an integer, as correctly verified in the main function.
Now, suppose I want to do the same with the second definition of f:
constexpr auto f() { return S{}, void(); }
In this case, the comma operator should intercept the form S{}, void().
Neither the following definition works (for obvious reasons):
constexpr int operator,(S, void) { return 42; }
Nor the one below (that would have worked in the previous case):
template<typename T> constexpr int operator,(S, T &&) { return 42; }
Is there any way to overload the comma operator so as to deal with S{}, void()?
Isn't it otherwise a lack in the standard, for it allows to use the comma operator that way, but doesn't give you the chance to overload the same operator (even if the standard mentions that overloaded functions involving S are allowed)?
Note: this question is made for the sake of curiosity. Please, avoid comments like do not do that or it is not good practice. I'm not planning to do that in production environments. Thank you.

The relevant clause for this is 13.3.1.2/9 [over.match.oper] in N4140:
If the operator is the operator ,, the unary operator &, or the operator ->, and there are no viable functions,
then the operator is assumed to be the built-in operator and interpreted according to Clause 5.
As void() is never a valid function argument (see 5.2.2/7 [expr.call]), there never is a viable function and thus the built-in , will be used.
So no, what you are trying to do is not possible.
In fact, writing an iterator loop like this
for(...; ++it1, (void)++it2)
is a standard way to prevent users from breaking your code by overloading , for their iterator types by enforcing the built-in operator , to be used. (Note that I am not saying you need to do this in your everyday code. It very much depends on its actual use. This is standard library level of paranoid.)
Regarding the standard clause you linked:
The meaning of the operators =, (unary) &, and , (comma), predefined for each type, can be changed for specific class and enumeration types by defining operator functions that implement these operators.
But such a function cannot be defined because, as I said above, void() is never a valid function argument.
Now whether or not this is an oversight/problem in the standard is open to debate.

Related

Casting to void to avoid use of overloaded user-defined Comma operator

I am learning about templates in C++, and came across an example where casting to void is used:
template<typename T>
auto func (T const& t) -> decltype( (void)(t.size()), T::size_type() )
{
return t.size();
}
In the explanation it is written that:
The cast of the expression to void is to avoid the possibility of a user-defined comma operator overloaded for the type of the expressions.
My question(s) is/are:
How can a cast to void be used to "avoid the possibility of a user-defined comma operator overloaded for the type of the expressions"? I mean, can anyone give any example where if we don't use void then this code would give an error? For example, let's say we have a class called SomeClass which has overloaded the comma operator. Now, can this become a problem if we don't use void?
Can static_cast be used in this case instead of a C style cast? For example, something like static_cast<void>(t.size()). I am reading examples that use C++17 features, and so I wonder why the author has used a C style cast in this case.
I have read What does casting to `void` really do?, from which I get the impression that if we use (void)x then this means to suppress compiler warnings, and also means "ignore the value of x". But then I can't understand the difference between the expression x and (void)x.
Consider this pathological type:
struct foo {
struct size_type {
bool operator,(size_type) { return false;}
};
size_type size() { return {};}
};
It does have a size_type and it does have a size() method. However, without the cast to void, the template does not deduce the right return type, because decltype( (t.size()), typename T::size_type() ) is bool :
#include <type_traits>
template<typename T>
auto func (T const& t) -> decltype( (t.size()), typename T::size_type() )
{
return t.size();
}
struct foo {
struct size_type {
bool operator,(size_type) { return false;}
};
size_type size() const { return {};}
};
int main()
{
func(foo{});
}
Results in error:
<source>:6:8: error: no viable conversion from returned value of type 'foo::size_type' to function return type 'decltype((t.size()) , typename foo::size_type())' (aka 'bool')
return t.size();
^~~~~~~~
<source>:20:4: note: in instantiation of function template specialization 'func<foo>' requested here
func(foo{});
^
1 error generated.
ASM generation compiler returned: 1
<source>:6:8: error: no viable conversion from returned value of type 'foo::size_type' to function return type 'decltype((t.size()) , typename foo::size_type())' (aka 'bool')
return t.size();
^~~~~~~~
<source>:20:4: note: in instantiation of function template specialization 'func<foo>' requested here
func(foo{});
^
A static_cast can be used. However, as nothing is actually being cast (it is an unevaluated context), the c-style cast does not do much harm. Note how by using the cast to void the user defined operator, is bypassed and the correct return type is deduced: https://godbolt.org/z/jozx1YGWr. This is because void, whatever uses the built-in operator, whose result is of same type as whatever. You cannot override void, whatever with a user-defined operator,; there is no syntax that works.
I suppose the code is merely to illustrate this one effect, because even with the cast to void one can make up other examples that fail (eg the template does not explicitly test for t.size() actually returning T::size_type).
See also the section on "Rarely overloaded operators" here:
The comma operator, operator,. Unlike the built-in version, the overloads do not sequence their left operand before the right one. (until C++17) Because this operator may be overloaded, generic libraries use expressions such as a,void(),b instead of a,b to sequence execution of expressions of user-defined types. The boost library uses operator, in boost.assign, boost.spirit, and other libraries. The database access library SOCI also overloads operator,.

Why is this user-defined conversion not done?

Consider:
template<typename T>
struct Prop
{
T value;
operator T() { return value; }
};
int main()
{
Prop<float> p1 { 5 };
Prop<std::vector<float>> p2 { { 1, 2, 3 } };
float f1 = p1; // Works fine
float f2_1_1 = p2.value[0]; // Works fine
float f2_1_2 = p2[0]; // Doesn't compile
return 0;
}
Why doesn't the line marked as such compile? Shouldn't it perform implicit conversion using the supplied conversion operator into std::vector<>, so that the [] can be found?
There are (many) other questions on this site that ask variations on this question, but I couldn't find one that I think applies here. Does it have to do with std::vector being a template?
Implicit conversions are not considered for objects of member function calls, including the subscript operator overload.
Consider the consequences if that were allowed: Every single time any undeclared member function is called like this, the compiler would have to figure out all the types that the object can be converted to (note that any other otherwise unrelated type could have a converting constructor), and check if that has declared the missing member function. Not to mention how confusing that would be for the reader of the code (the conversion might be obvious in your conversion operator case, but not in the converting constructor case, and as far as I know, they are not otherwise treated differently).
So is there a notationally convenient way to get Prop to behave the way I want
You would have to define a member function for each of the member function of vector that you want to pass pass through transparently. Here is the subscript operator as an example:
auto operator[](std::size_t pos) {
return value[pos];
}
auto operator[](std::size_t pos) const {
return value[pos];
}
The problem is of course that all wrapped member functions must be explicitly declared. Another problem is arguments whose type depend on T. For example, vector::operator[] uses vector::size_type, which might not be defined for all T that you might use (certainly not for float). Here we make a compromise and use std::size_t.
A less laborious way of creating such "transparent" wrapper is inheritance. A publicly inheriting template would automatically have all the member functions of the parent and be implicitly convertible to it and can be referred by pointers and references of parent type. However, the transparency of such approach is a bit problematic mainly because ~vector is not virtual.
Private inheritance allows same wrapping as your member approach, but with much nicer syntax:
template<typename T>
struct Prop : private T
{
using T::operator[];
using T::T;
};
Note that inheritance approach prevents you from using fundamental types as T. Also, it makes the implicit conversion impossible (even with conversion operator) , so you cannot use Prop as T in free functions that expect T.
PS. Note that your conversion operator returns a value, so the vector must be copied, which may be undesirable. Consider providing reference versions instead:
operator T&&()&& { return *this; }
operator T&()& { return *this; }
operator const T&() const& { return *this; }
the same way that any of the
p2.size();
p2.begin();
p2.push_back(24);
// etc.
don't make sense to compile
also
p2[0];
which is equivalent with
p2.operator[](0);
doesn't make sense to compile\
If you want this behavior (i.e. for Prop<T> to borrow T members) there is a C++ proposal by Bjarne to add the dot operator to the language. I would work the same way that operator -> works for smart pointers. AFAIR it had a lot of controversy and so I wouldn't hold my breath for it.
A bit of background for the operator dot proposal—Bjarne Stroustrup
Operator Dot (R3) - Bjarne Stroustrup, Gabriel Dos Rei
Smart References through Delegation: An Alternative to N4477's Operator Dot - Hubert Tong, Faisal Vali
Alternatives to operator dot - Bjarne Stroustrup

C++ Assign to implicitly converted lvalue

Consider this snippet of C++ code:
struct Foo {
float value;
operator float& () {
return this->value;
}
};
int main() {
Foo foo;
foo=1.0f; //Doesn't compile, foo isn't implicitly converted to a float&
return 0;
}
Why doesn't this compile? Is there a specific reason this wasn't included in the C++ standard? Or an equivalent does indeed exist and I'm just using it wrong?
For pretty much all other operators, your conversion operator would do exactly what you want, and it would continue to do exactly what you want even if you add custom operators.
struct Foo {
float value;
operator float& () { return this->value; }
Foo &operator+=(Foo);
};
int main() {
Foo foo {};
foo+=1.0; // compiles and works
// equivalent to foo.operator float&()+=1.0;
}
However, = is special, the rules for = are different compared to most other operators. As identified by T.C.:
13.3.1.2 Operators in expressions [over.match.oper]
4 For the built-in assignment operators, conversions of the left operand are restricted as follows:
(4.1) -- no temporaries are introduced to hold the left operand, and
(4.2) -- no user-defined conversions are applied to the left operand to achieve a type match with the left-most parameter of a built-in candidate.
Together with the fact that any custom operator= is not allowed to be defined as a global function, this makes sure that foo=bar; where foo is a class type always means foo.operator=(bar);, nothing else.
The fact that this one operator is singled out doesn't explain the reason, but does make it quite clear that it's an intentional decision, and making sure that foo=bar; always means foo.operator=(bar);, nothing else, by itself already seems like a valid reason.
Implicit conversion is only done in the following cases:
Implicit conversions are performed whenever an expression of some type
T1 is used in context that does not accept that type, but accepts some
other type T2; in particular:
when the expression is used as the argument when calling a function that is declared with T2 as parameter;
when the expression is used as an operand with an operator that expects T2;
when initializing a new object of type T2, including return statement in a function returning T2;
when the expression is used in a switch statement (T2 is integral type);
when the expression is used in an if statement or a loop (T2 is bool).
None of those is the case here. Instead the compiler is trying to find a suitable operator= to work with a double. To have this compile you need to overload that operator ( you actually want a float as seen in the code ):
Foo& operator=(float other)
{
value = f;
return *this;
}
And change your assignment to foo = 1.0f;
Your conversion function would work with for example:
float f = foo;

Overloading conversion operator template

Consider the following simple example
struct C
{
template <typename T> operator T () {return 0.5;}
operator int () {return 1;}
operator bool () {return false;}
};
int main ()
{
C c;
double x = c;
std::cout << x << std::endl;
}
When compiled with Clang, it gives the following error
test.cpp:11:12: error: conversion from 'C' to 'double' is ambiguous
double x = c;
^ ~
test.cpp:4:5: note: candidate function
operator int () {return 1;}
^
test.cpp:5:5: note: candidate function
operator bool () {return false;}
^
test.cpp:3:27: note: candidate function [with T = double]
template <typename T> operator T () {return 0.5;}
^
1 error generated.
Other compilers generate similar errors, e.g., GCC and Intel iclc
If I remove operator int and operator bool. It compiles fine and work as expected. If only remove one of them, that is keep the template operator and say operator int, then the non-template version is always chosen.
My understanding is that only when the template and non-template overloaded functions are equal in the sense that they are both perfect match or both require the same conversion sequence, the non-template version will be preferred. However in this case, it appears that the compiler does not see the operator template as a perfect match. And when both the bool and int overloading are present, then naturally it considers they are ambiguous.
In summary, my question is that why the operator template is not considered a perfect match in this case?
Let's break this down into two different problems:
1. Why does this generate a compiler error?
struct C
{
operator bool () {return false;}
operator int () {return 1;}
};
As both int and bool can be implicitly converted to double, the compiler can not know which function it should use. There are two functions which it could use and neither one takes precedence over the other one.
2. Why isn't the templated version a perfect match?
struct C
{
template <typename T> operator T () {return 0.5;}
operator int () {return 1;}
};
Why is operator int() called when requesting a double?
The non-template function is called because a non-template function takes precedence in overload resolution. (Overloading function templates)
EDIT:
I was wrong! As Yan Zhou mentioned in his comment, and as it is stated in the link I provided, a perfect match in the templated function takes precedence over the non-templated function.
I tested your code (compiled with g++ 4.7.2), and it worked as expected: it returned 0.5, in other words, the templated function was used!
EDIT2:
I now tried with clang and I can reproduce the behaviour you described. As it works correctly in gcc, this seems to be a bug in clang.
This is interesting. There are two ways to read a critical part of section 13.3.3. The original example should definitely call the function template, but the version where one of the non-templates is removed might be argued to be ambiguous.
13.3.3:
A viable function F1 is defined to be a better function than another viable function F2 if for all arguments i, ICS_i(F1) is not a worse conversion sequence than ICS_i(F2), and then
for some argument j, ICS_j(F1) is a better conversion sequence than ICS_j(F2), or, if not that,
the context is an initialization by user-defined conversion (see 8.5, 13.3.1.5, and 13.3.1.6) and the standard conversion sequence from the return type of F1 to the destination type (i.e., the type of the entity being initialized) is a better conversion sequence than the standard conversion sequence from the return type of F2 to the destination type, or, if not that,
F1 is a non-template function and F2 is a function template specialization, or, if not that,
F1 and F2 are function template specializations, and the function template for F1 is more specialized than the template for F2 according to the partial ordering rules described in 14.5.6.2.
If there is exactly one viable function that is a better function than all other viable functions, then it is the one selected by overload resolution; otherwise the call is ill-formed.
In the example, clang correctly identifies the set of three viable candidate functions:
C::operator int()
C::operator bool()
C::operator double<double>()
The third is a function template specialization. (I don't think the syntax above is legal, but you get the idea: at this point of overload resolution it's not treated as a template, but as a specialization with a definite function type.)
The only Implicit Conversion Sequence on arguments here (ICS1) is the exact match "lvalue C" to "C&" on the implicit parameter, so that won't make a difference.
This example is exactly the situation described in the second bullet, so the function returning double is clearly better than the other two.
Here's where it gets weird: By a very literal reading, operator int is also better than the template specialization, because of the third bullet. "Wait a minute, shouldn't 'better than' be antisymmetric? How can you say F1 is better than F2 AND F2 is better than F1?" Unfortunately, the Standard doesn't explicitly say anything of the sort. "Doesn't the second bullet take priority over the third bullet because of the 'if not that' phrase?" Yes, for constant F1 and F2. But the Standard doesn't say that satisfying the second bullet for (F1,F2) makes the third bullet for (F2,F1) not applicable.
Of course, since operator int is not better than operator bool and vice versa, there is still "exactly one viable function that is a better function than all other viable functions".
I'm not exactly endorsing this weird reading, except maybe to report it as a Standard defect. Going with that would have bizarre consequences (like removing an overload which was not the best from this example changes a program from well-formed to ambiguous!). I think the intent is for the second bullet to be considered both ways before the third bullet is considered at all.
Which would mean the function template should be selected by overload resolution, and this is a clang bug.

overloading non-member conversion to bool operator

I am trying to write bool-conversion operator for std::bitset
I tried:
template<size_t size>
operator bool(std::bitset<size> & b)
{
return b.any();
}
but I got
error C2801: 'mynamespace::operator bool' must be a non-static member
from my visual-studio.
But when I look up C2801 explanation it says nothing about conversion operators (only about =, ->, [],())
So, is it possible to somehow write "Conversion std::bitset to bool operator?"
(I can not call b.any() in my if-statements, because the same code must run when std::bitset is replaced with unsigned or something
typedef std::bitset<x> Bitset;
//typedef unsigned Bitset;
so the ideal syntax will be like:
Bitset b = whatewer;
if(b)
doStuff();
)
If this overloading is not possible, what is the recommended workaround?
so far I use it like:
if(b == Bitset(0))
doStuff();
but I dont like it.
Thank you
As the error message says, the conversion operator must be a non-static member of a class. That is true.
I can not call b.any() in my if-statements, because the same code must run when std::bitset is replaced with unsigned or something.
If that is your problem, then you can use function overload, and call it passing the argument which will return a boolean value:
template<typename T>
bool to_bool(T const & b)
{
return b; //implicit conversion (if allowed) for all other types
}
template<size_t N>
bool to_bool(std::bitset<N> const & b)
{
return b.any();
}
then use it as:
if (to_bool(whatever))
{
}
It will call the correct overload. If the type of whatever is std::bitset<N> then the second overloaded function will be called, or else the first one will be called.
§12.3.2/1: "A member function of a class X with a name of the form [...] specifies a conversion from X to the type specified..." (C++11 uses the same section number and nearly the same wording, adding only that the function takes no parameters).
The other possible way to define a conversion is a constructor (§12.3.1), which is obviously a class member as well.
In short, yes, conversions must always be defined as member functions.
One way to do what you want would be to write a wrapper around std::bitset that provides the conversion you care about:
template <int size>
class mybitest {
std::bitset<size> bits;
public:
operator bool() { return bits.any(); }
}
But if you decide to do that, you'll need to write forwarding functions for essentially all the pieces of bitset you're using (ctors, assignment, etc.)
The standard is a bit unclear on this (12.3.2):
A member function of a class X having no parameters with a name of the form [...] specifies a conversion from X to the type specified by the conversion-type-id. Such functions are called conversion functions. No return type can be specified. If a conversion function is a member function, the type of the conversion function (8.3.5) is “function taking no parameter returning conversion-type-id”.
The first sentence seems to imply that only member functions can be conversion functions, but I'm not sure what the purpose of the conditional "if a conversion function is a member function" is.
I'd take the first sentence as binding and conclude that a conversion function must be a member function.
in case this helps somebody, you can actually provide a not operator instead
template<size_t size>
operator !(std::bitset<size> & b)
{
return !b.any();
}
and use it like so using the !! idiom:
if (!!whatever)
{
}
still not ideal, but a bit closer I think.