C++ template function are expected to throw exception? - c++

I was searching for reasons that prevent my compiler (intel 15) to vectorize my code. The reasons were : loop with early exits cannot be vectorized unless it meets search loop idiom criteria although I did not have any thing like break or exit(0) in it.
So I checked all my function with the noexcept operator and it seems that all my template function are likely to throw exception (the operator returns false all the time) althgough I have no try/catch/throw in it.
For this reason, I specify the noexcept on my function and now it works but I don't understand why the compiler say that they are likely to throw something ? How can I specify that, by default, they won't throw anything ??
Example:
/*.h*/
template <typename T>
void A(T& a){} /*does nothing*/
/*.cpp*/
#include ".h"
#include "MyClass.h"
int main(){
MyClass C;
noexcept(A(C)); /*Return false !!*/
}
Thanks.

If you look at the documentation for the noexcept operator (emphasis mine)
The result is false if the expression contains at least one of the following potentially evaluated constructs:
call to any type of function that does not have non-throwing exception specification, unless it is a constant expression.
throw expression.
dynamic_cast expression when the target type is a reference type, and conversion needs a run time check
typeid expression when argument type is polymorphic class type
In all other cases the result is true.
So as you noticed, the way to have your template function evaluate noexcept as true is simply to declare it as such
template <typename T>
void A(T& a) noexcept
{
}

Related

Implement assert without preprocessor in C++20

C++ knows assert() which allows runtime checks that compile to nothing in dependence of NDEBUG.
I would like to replace that macro using compiler-code and avoiding the preprocessor. I need to do the following:
Break or terminate if expressions evaluates to false
Log the code line where the assert is called
Discard the check and the passed expression for NDEBUG builds
Breaking/terminating the application is easy.
In C++20 there is std::experimental::source_location which I can use to get the code location of the assertion.
A compile time conditional could be done using requires or constexpr if
However I do not know how I could avoid the evaluation of the expression. When implementing myAssert(expression) as a function, I need to pass the expression result as a function argument which means it is evaluated anyway, even if the parameter is not used inside the function.
Is there a way to solve this in C++20?
EDIT: A templated example:
template <typename T> requires (gDebug)
void assertTrue(const T& pResult, const std::experimental::source_location& pLocation) noexcept
{
if (!static_cast<bool>(pResult))
{
// error handling
}
}
template <typename T> requires (!gDebug)
void assertTrue(const T&) noexcept
{
}
I suppose you are talking about the case when you disabled debugging and you want the function to be a noop. I see 2 options:
You can use a macro. Macros can be misused, but they have their place and "passing an expression" without evaluating it is a case for a macro.
Alternatively, pass a callable that returns the result you want to assert for. Only call it when gDebug == True:
template <typename F> requires (gDebug)
void assertTrue(const F& f, const std::experimental::source_location& pLocation) noexcept
{
if (!static_cast<bool>(f()))
{
// error handling
}
}
Though this will make the call rather verbose. For example one that fails always:
assertTrue( [](){ return false; });

While trying to use SFINAE to disable functions, have I created undefined behavior?

I am trying to use SFINAE to disable certain functions of a class based on some non-templated enum arguments.
The following code does NOT compile with gcc, but appears to compile and work as expected when using the msvc compiler.
#include <iostream>
#include <type_traits>
enum class B { VARIANT1, VARIANT2 };
template<B B_VAL>
struct A {
template<class = std::enable_if_t<B_VAL == B::VARIANT1>>
void func1() {
std::cout<<"VARIANT1"<<std::endl;
}
template<class = std::enable_if_t<B_VAL == B::VARIANT2>>
void func2() {
std::cout<<"VARIANT2"<<std::endl;
}
};
int main()
{
A<B::VARIANT1> a;
a.func1();
}
The expected (and msvcs) behavior is that calling a function whose enable_if_t condition equates to false results in a compile time error, or the removal of the function candidate for overload resolution if an overloaded function was present in the example. In all other cases, the code should compile normally.
gcc on the other hand tells me that it can't find a type named "type" in "struct std::enable_if<false, void>" for the enable_if_t in the template of func2, which makes perfect sense as the member named "type" is only present in enable_if if the condition equates to true. But shouldn't this be the desired behavior for the SFINAE functionality and shouldn't the compiler ignore func2, as it is never called?
I now have three questions:
As the two compilers produce different behavior, is it undefined and if yes, which parts/statements?
Is SFINAE suited to achieve my goal, or have I misunderstood its use case?
Would I be better off by using static asserts as alternative?
I am sorry if this question is a duplicate of this one, but I don't think that the answers there provided much help with my problem.
GCC is right. And it's because you aren't using SFINAE for your function. It may appear that you do because you employ utilities for SFINAE from the standard library, but there is a crucial ingredient missing here.
The 'S' in "SFINAE" stands for substitution. The substitution of template arguments into the parameters of a template we are trying to instantiate. Now, the template in question is func2. And for SFINAE to work, it is func2's argument that must fail to be substituted for its parameters. But here
std::enable_if_t<B_VAL == B::VARIANT2>
There is no parameter of func2 in use. It doesn't depend on anything that happens during substitution into func2. It's just an invalid type, completely independent of an attempt to actually instantiate func2.
It's not hard to fix though
template<B B_VAL_ = B_VAL, class = std::enable_if_t<B_VAL_ == B::VARIANT1>>
void func1() {
std::cout<<"VARIANT1"<<std::endl;
}
template<B B_VAL_ = B_VAL, class = std::enable_if_t<B_VAL_ == B::VARIANT2>>
void func2() {
std::cout<<"VARIANT2"<<std::endl;
}
Now, the check is against the substitution into the correct template.

Is it possible to test whether something is constexpr? [duplicate]

Is it possible to produce a compile-time boolean value based on whether or not a C++11 expression is a constant expression (i.e. constexpr) in C++11? A few questions on SO relate to this, but I don't see a straight answer anywhere.
I once wrote it (EDIT: see below for limitations and explanations). From https://stackoverflow.com/a/10287598/34509 :
template<typename T>
constexpr typename remove_reference<T>::type makeprval(T && t) {
return t;
}
#define isprvalconstexpr(e) noexcept(makeprval(e))
However there are many kinds of constant expressions. The above answer detects prvalue constant expressions.
Explanation
The noexcept(e) expression gives false iff e contains
a potentially evaluated call to a function that does not have a non-throwing exception-specification unless the call is a constant expression,
a potentially evaluated throw expression,
a potentially evaluated throwable form of dynamic_cast or typeid.
Note that the function template makeprval is not declared noexcept, so the call needs to be a constant expression for the first bullet not to apply, and this is what we abuse. We need the other bullets to not apply aswell, but thanksfully, both a throw and a throwable dynamic_cast or typeid aren't allowed in constant expressions aswell, so this is fine.
Limitations
Unfortunately there is a subtle limitation, which may or may not matter for you. The notion of "potentially evaluated" is much more conservative than the limits of what constant expressions apply. So the above noexcept may give false negatives. It will report that some expressions aren't prvalue constant expressions, even though they are. Example:
constexpr int a = (0 ? throw "fooled!" : 42);
constexpr bool atest = isprvalconstexpr((0 ? throw "fooled!" : 42));
In the above atest is false, even though the initialization of a succeeded. That is because for being a constant expression, it suffices that the "evil" non-constant sub-expressions are "never evaluated", even though those evil sub-expressions are potentially-evaluated, formally.
As of 2017, is_constexpr is not possible in C++11. That sounds like an odd thing to say, so let me explain a bit of the history.
First, we added this feature to resolve a defect: http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1129
Johannes Schaub - litb posted a constexpr detection macro that relied on the provision that constant expressions are implicitly noexcept. This worked in C++11, but was never implemented by at least some compilers (for instance, clang). Then, as part of C++17, we evaluated Removing Deprecated Exception Specifications from C++17. As a side-effect of that wording, we accidentally removed that provision. When the Core Working Group discussed adding the provision back in, they realized that there were some serious problems with doing so. You can see the full details in the LLVM bug report. So rather than adding it back in, we decided to consider it a defect against all versions of standard and retroactively removed it.
The effect of this is that there is, to my knowledge, no way to detect whether an expression is usable as a constant expression.
Yes, this is possible. One way to do it (which is valid even with the recent noexcept changes) is to take advantage of the C++11 narrowing conversion rules:
A narrowing conversion is an implicit conversion [...] from an integer type or unscoped enumeration type to an integer type that cannot represent all the values of the original type, except where the source is a constant expression whose value after integral promotions will fit into the target type.
(emphasis mine). List initialization generally disallows narrowing conversions, and when combined with SFINAE we can build gadgets for detecting whether an arbitrary expression is a constant expression:
// p() here could be anything
template<int (*p)()> std::true_type is_constexpr_impl(decltype(int{(p(), 0U)}));
template<int (*p)()> std::false_type is_constexpr_impl(...);
template<int (*p)()> using is_constexpr = decltype(is_constexpr_impl<p>(0));
constexpr int f() { return 0; }
int g() { return 0; }
static_assert(is_constexpr<f>());
static_assert(!is_constexpr<g>());
Live demonstration.
The key here is that int{(expr, 0U)} contains a narrowing conversion from unsigned int to int (and thus is ill-formed), unless expr is a constant expression, in which case the entire expression (expr, 0U) is a constant expression whose evaluated value fits into the type int.
The following is an implementation of is_constexpr for functions, not for arbitrary expressions, for C++11 and C++17. It requires the arguments to the function you want to test to be default constructible, though.
#include <type_traits>
struct A {}; // don't make it too easy, use a UDT
A f1(A a) { return a; } // is_constexpr -> false
constexpr A f2(A a) { return a; } // is_constexpr -> true
// The following turns anything (in our case a value of A) into an int.
// This is necessary because non-type template arguments must be integral
// (likely to change with C++20).
template <class T> constexpr int make_int(T &&) { return 0; }
// Helper to turn some function type (e.g. int(float)) into a function
// pointer type (e.g. int (*)(float)).
template <class T> struct signature_from;
template <class R, class... Args> struct signature_from<R(Args...)> {
using type = R(*)(Args...);
};
// See std::void_t for the idea. This does it for ints instead of types.
template <int...> using void_from_int = void;
// The fallback case: F is not a function pointer to a constexpr function
template <class T, typename signature_from<T>::type F, class = void_from_int<>>
struct is_constexpr {
static constexpr bool value = false;
};
// If void_from_int<make_int(F(Args()...))> doesn't lead to a substitution
// failure, then this is the preferred specialization. In that case F must
// be a function pointer to a constexpr function. If it is not, it could
// not be used in a template argument.
template <class R, class... Args, typename signature_from<R(Args...)>::type F>
struct is_constexpr<R(Args...), F, void_from_int<make_int(F(Args()...))>>
{
static constexpr bool value = true;
};
// proof that it works:
static_assert(!is_constexpr<A(A), f1>::value, "");
static_assert( is_constexpr<A(A), f2>::value, "");
#if __cplusplus >= 201703
// with C++17 the type of the function can be deduced:
template<auto F> struct is_constexpr2 : is_constexpr<std::remove_pointer_t<decltype(F)>, F> {};
static_assert(!is_constexpr2<f1>::value, "");
static_assert( is_constexpr2<f2>::value, "");
#endif
See it in action at https://godbolt.org/g/rdeQme.
C++20 added std::is_constant_evaluated()
This allows checking if a certain expression is a constant evaluated expression, i.e. being evaluated at compile time.
Usage example:
constexpr int foo(int num) {
// below is true in case the condition is being evaluated at compile time
// side note, using: if constexpr (std::is_constant_evaluated())
// would be evaluated always to true, so you should use a simple if!
if (std::is_constant_evaluated()) {
return foo_compiletime(num);
}
else {
return foo_runtime(num);
}
}
int main() {
constexpr auto t1 = foo(6); // reaches foo_compiletime
const auto t2 = foo(6); // reaches foo_compiletime
int n = rand() % 10;
const auto t3 = foo(n); // reaches foo_runtime
auto t4 = foo(6); // unfortunately, reaches foo_runtime
}
The last call in the example above would reach foo_runtime, since the call is not within a constant expression context (the result is not being used as a constant expression, see also this SO answer).
This may lead to undesired pessimization, compared to the case of leaving the decision to the user, who may call:
auto t4 = foo_compiletime(6);
And the compiler is allowed to perform the operations inside foo_compiletime at compile time, if it is declared as constexpr function, or would be obliged to do that if it is declared consteval. However, once we leave the decision to the compiler, we will reach foo_runtime, unless we explicitly direct the compiler to go for foo_compiletime, by taking the result into a const, constexpr or constinit variable. Which then, in a way, omits the value of having one function for both scenarios, if the user is required to help the compiler peek the right path.
Another possible option for the call to be optimized, is:
constexpr auto temp = foo(6); // foo_compiletime
auto t4 = temp;
But again, we require the user to be aware of the inner behavior of foo, which is not exactly what we want to achieve.
See the pessimization in this code.
See more on that in this great blog post on the subject.

How does this implementation of std::is_class work?

I'm trying to understand the implementation of std::is_class. I've copied some possible implementations and compiled them, hoping to figure out how they work. That done, I find that all the computations are done during compilation (as I should have figured out sooner, looking back), so gdb can give me no more detail on what exactly is going on.
The implementation I'm struggling to understand is this one:
template<class T, T v>
struct integral_constant{
static constexpr T value = v;
typedef T value_type;
typedef integral_constant type;
constexpr operator value_type() const noexcept {
return value;
}
};
namespace detail {
template <class T> char test(int T::*); //this line
struct two{
char c[2];
};
template <class T> two test(...); //this line
}
//Not concerned about the is_union<T> implementation right now
template <class T>
struct is_class : std::integral_constant<bool, sizeof(detail::test<T>(0))==1
&& !std::is_union<T>::value> {};
I'm having trouble with the two commented lines. This first line:
template<class T> char test(int T::*);
What does the T::* mean? Also, is this not a function declaration? It looks like one, yet this compiles without defining a function body.
The second line I want to understand is:
template<class T> two test(...);
Once again, is this not a function declaration with no body ever defined? Also what does the ellipsis mean in this context? I thought an ellipsis as a function argument required one defined argument before the ...?
I would like to understand what this code is doing. I know I can just use the already implemented functions from the standard library, but I want to understand how they work.
References:
std::is_class
std::integral_constant
What you are looking at is some programming technologie called "SFINAE" which stands for "Substitution failure is not an error". The basic idea is this:
namespace detail {
template <class T> char test(int T::*); //this line
struct two{
char c[2];
};
template <class T> two test(...); //this line
}
This namespace provides 2 overloads for test(). Both are templates, resolved at compile time. The first one takes a int T::* as argument. It is called a Member-Pointer and is a pointer to an int, but to an int thats a member of the class T. This is only a valid expression, if T is a class.
The second one is taking any number of arguments, which is valid in any case.
So how is it used?
sizeof(detail::test<T>(0))==1
Ok, we pass the function a 0 - this can be a pointer and especially a member-pointer - no information gained which overload to use from this.
So if T is a class, then we could use both the T::* and the ... overload here - and since the T::* overload is the more specific one here, it is used.
But if T is not a class, then we cant have something like T::* and the overload is ill-formed. But its a failure that happened during template-parameter substitution. And since "substitution failures are not an error" the compiler will silently ignore this overload.
Afterwards is the sizeof() applied. Noticed the different return types? So depending on T the compiler chooses the right overload and therefore the right return type, resulting in a size of either sizeof(char) or sizeof(char[2]).
And finally, since we only use the size of this function and never actually call it, we dont need an implementation.
Part of what is confusing you, which isn't explained by the other answers so far, is that the test functions are never actually called. The fact they have no definitions doesn't matter if you don't call them. As you realised, the whole thing happens at compile time, without running any code.
The expression sizeof(detail::test<T>(0)) uses the sizeof operator on a function call expression. The operand of sizeof is an unevaluated context, which means that the compiler doesn't actually execute that code (i.e. evaluate it to determine the result). It isn't necessary to call that function in order to know the sizeof what the result would be if you called it. To know the size of the result the compiler only needs to see the declarations of the various test functions (to know their return types) and then to perform overload resolution to see which one would be called, and so to find what the sizeof the result would be.
The rest of the puzzle is that the unevaluated function call detail::test<T>(0) determines whether T can be used to form a pointer-to-member type int T::*, which is only possible if T is a class type (because non-classes can't have members, and so can't have pointers to their members). If T is a class then the first test overload can be called, otherwise the second overload gets called. The second overload uses a printf-style ... parameter list, meaning it accepts anything, but is also considered a worse match than any other viable function (otherwise functions using ... would be too "greedy" and get called all the time, even if there's a more specific function t hat matches the arguments exactly). In this code the ... function is a fallback for "if nothing else matches, call this function", so if T isn't a class type the fallback is used.
It doesn't matter if the class type really has a member variable of type int, it is valid to form the type int T::* anyway for any class (you just couldn't make that pointer-to-member refer to any member if the type doesn't have an int member).
The std::is_class type trait is expressed through a compiler intrinsic (called __is_class on most popular compilers), and it cannot be implemented in "normal" C++.
Those manual C++ implementations of std::is_class can be used in educational purposes, but not in a real production code. Otherwise bad things might happen with forward-declared types (for which std::is_class should work correctly as well).
Here's an example that can be reproduced on any msvc x64 compiler.
Suppose I have written my own implementation of is_class:
namespace detail
{
template<typename T>
constexpr char test_my_bad_is_class_call(int T::*) { return {}; }
struct two { char _[2]; };
template<typename T>
constexpr two test_my_bad_is_class_call(...) { return {}; }
}
template<typename T>
struct my_bad_is_class
: std::bool_constant<sizeof(detail::test_my_bad_is_class_call<T>(nullptr)) == 1>
{
};
Let's try it:
class Test
{
};
static_assert(my_bad_is_class<Test>::value == true);
static_assert(my_bad_is_class<const Test>::value == true);
static_assert(my_bad_is_class<Test&>::value == false);
static_assert(my_bad_is_class<Test*>::value == false);
static_assert(my_bad_is_class<int>::value == false);
static_assert(my_bad_is_class<void>::value == false);
As long as the type T is fully defined by the moment my_bad_is_class is applied to it for the first time, everything will be okay. And the size of its member function pointer will remain what it should be:
// 8 is the default for such simple classes on msvc x64
static_assert(sizeof(void(Test::*)()) == 8);
However, things become quite "interesting" if we use our custom type trait with a forward-declared (and not yet defined) type:
class ProblemTest;
The following line implicitly requests the type int ProblemTest::* for a forward-declared class, definition of which cannot be seen by the compiler right now.
static_assert(my_bad_is_class<ProblemTest>::value == true);
This compiles, but, unexpectedly, breaks the size of a member function pointer.
It seems like the compiler attempts to "instantiate" (similarly to how templates are instantiated) the size of a pointer to ProblemTest's member function in the same moment that we request the type int ProblemTest::* within our my_bad_is_class implementation. And, currently, the compiler cannot know what it should be, thus it has no choice but to assume the largest possible size.
class ProblemTest // definition
{
};
// 24 BYTES INSTEAD OF 8, CARL!
static_assert(sizeof(void(ProblemTest::*)()) == 24);
The size of a member function pointer was trippled! And it cannot be shrunk back even after the definition of class ProblemTest has been seen by the compiler.
If you work with some third party libraries that rely on particular sizes of member function pointers on your compiler (e.g., the famous FastDelegate by Don Clugston), such unexpected size changes caused by some call to a type trait might be a real pain. Primarily because type trait invocations are not supposed to modify anything, yet, in this particular case, they do -- and this is extremely unexpected even for an experienced developer.
On the other hand, had we implemented our is_class using the __is_class intrinsic, everything would have been OK:
template<typename T>
struct my_good_is_class
: std::bool_constant<__is_class(T)>
{
};
class ProblemTest;
static_assert(my_good_is_class<ProblemTest>::value == true);
class ProblemTest
{
};
static_assert(sizeof(void(ProblemTest::*)()) == 8);
Invocation of my_good_is_class<ProblemTest> does not break any sizes in this case.
So, my advice is to rely on the compiler intrinsics when implementing your custom type traits like is_class wherever possible. That is, if you have a good reason to implement such type traits manually at all.
What does the T::* mean? Also, is this not a function declaration? It looks like one, yet this compiles without defining a function body.
The int T::* is a pointer to member object. It can be used as follows:
struct T { int x; }
int main() {
int T::* ptr = &T::x;
T a {123};
a.*ptr = 0;
}
Once again, is this not a function declaration with no body ever defined? Also what does the ellipsis mean in this context?
In the other line:
template<class T> two test(...);
the ellipsis is a C construct to define that a function takes any number of arguments.
I would like to understand what this code is doing.
Basically it's checking if a specific type is a struct or a class by checking if 0 can be interpreted as a member pointer (in which case T is a class type).
Specifically, in this code:
namespace detail {
template <class T> char test(int T::*);
struct two{
char c[2];
};
template <class T> two test(...);
}
you have two overloads:
one that is matched only when a T is a class type (in which case this one is the best match and "wins" over the second one)
on that is matched every time
In the first the sizeof the result yields 1 (the return type of the function is char), the other yields 2 (a struct containing 2 chars).
The boolean value checked is then:
sizeof(detail::test<T>(0)) == 1 && !std::is_union<T>::value
which means: return true only if the integral constant 0 can be interpreted as a pointer to member of type T (in which case it's a class type), but it's not a union (which is also a possible class type).
Test is an overloaded function that either takes a pointer to member in T or anything. C++ requires that the best match be used. So if T is a class type it can have a member in it...then that version is selected and the size of its return is 1. If T is not a class type then T::* make zero sense so that version of the function is filtered out by SFINAE and won't be there. The anything version is used and it's return type size is not 1. Thus checking the size of the return of calling that function results in a decision whether the type might have members...only thing left is making sure it's not a union to decide if it's a class or not.
Here is standard wording:
[expr.sizeof]:
The sizeof operator yields the number of bytes occupied by a non-potentially-overlapping object of the type of its operand.
The operand is either an expression, which is an unevaluated operand
([expr.prop])......
2. [expr.prop]:
In some contexts, unevaluated operands appear ([expr.prim.req], [expr.typeid], [expr.sizeof], [expr.unary.noexcept], [dcl.type.simple], [temp]).
An unevaluated operand is not evaluated.
3. [temp.fct.spec]:
[Note: Type deduction may fail for the following reasons:
...
(11.7) Attempting to create “pointer to member of T” when T is not a class type.
[ Example:
template <class T> int f(int T::*);
int i = f<int>(0);
— end example
]
As above shows, it is well-defined in standard :-)
4. [dcl.meaning]:
[Example:
struct X {
void f(int);
int a;
};
struct Y;
int X::* pmi = &X::a;
void (X::* pmf)(int) = &X::f;
double X::* pmd;
char Y::* pmc;
declares pmi, pmf, pmd and pmc to be a pointer to a member of X of type int, a pointer to a member of X of type void(int), a pointer to a member ofX of type double and a pointer to a member of Y of type char respectively.The declaration of pmd is well-formed even though X has no members of type double. Similarly, the declaration of pmc is well-formed even though Y is an incomplete type.

Is is_constexpr possible in C++11?

Is it possible to produce a compile-time boolean value based on whether or not a C++11 expression is a constant expression (i.e. constexpr) in C++11? A few questions on SO relate to this, but I don't see a straight answer anywhere.
I once wrote it (EDIT: see below for limitations and explanations). From https://stackoverflow.com/a/10287598/34509 :
template<typename T>
constexpr typename remove_reference<T>::type makeprval(T && t) {
return t;
}
#define isprvalconstexpr(e) noexcept(makeprval(e))
However there are many kinds of constant expressions. The above answer detects prvalue constant expressions.
Explanation
The noexcept(e) expression gives false iff e contains
a potentially evaluated call to a function that does not have a non-throwing exception-specification unless the call is a constant expression,
a potentially evaluated throw expression,
a potentially evaluated throwable form of dynamic_cast or typeid.
Note that the function template makeprval is not declared noexcept, so the call needs to be a constant expression for the first bullet not to apply, and this is what we abuse. We need the other bullets to not apply aswell, but thanksfully, both a throw and a throwable dynamic_cast or typeid aren't allowed in constant expressions aswell, so this is fine.
Limitations
Unfortunately there is a subtle limitation, which may or may not matter for you. The notion of "potentially evaluated" is much more conservative than the limits of what constant expressions apply. So the above noexcept may give false negatives. It will report that some expressions aren't prvalue constant expressions, even though they are. Example:
constexpr int a = (0 ? throw "fooled!" : 42);
constexpr bool atest = isprvalconstexpr((0 ? throw "fooled!" : 42));
In the above atest is false, even though the initialization of a succeeded. That is because for being a constant expression, it suffices that the "evil" non-constant sub-expressions are "never evaluated", even though those evil sub-expressions are potentially-evaluated, formally.
As of 2017, is_constexpr is not possible in C++11. That sounds like an odd thing to say, so let me explain a bit of the history.
First, we added this feature to resolve a defect: http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1129
Johannes Schaub - litb posted a constexpr detection macro that relied on the provision that constant expressions are implicitly noexcept. This worked in C++11, but was never implemented by at least some compilers (for instance, clang). Then, as part of C++17, we evaluated Removing Deprecated Exception Specifications from C++17. As a side-effect of that wording, we accidentally removed that provision. When the Core Working Group discussed adding the provision back in, they realized that there were some serious problems with doing so. You can see the full details in the LLVM bug report. So rather than adding it back in, we decided to consider it a defect against all versions of standard and retroactively removed it.
The effect of this is that there is, to my knowledge, no way to detect whether an expression is usable as a constant expression.
Yes, this is possible. One way to do it (which is valid even with the recent noexcept changes) is to take advantage of the C++11 narrowing conversion rules:
A narrowing conversion is an implicit conversion [...] from an integer type or unscoped enumeration type to an integer type that cannot represent all the values of the original type, except where the source is a constant expression whose value after integral promotions will fit into the target type.
(emphasis mine). List initialization generally disallows narrowing conversions, and when combined with SFINAE we can build gadgets for detecting whether an arbitrary expression is a constant expression:
// p() here could be anything
template<int (*p)()> std::true_type is_constexpr_impl(decltype(int{(p(), 0U)}));
template<int (*p)()> std::false_type is_constexpr_impl(...);
template<int (*p)()> using is_constexpr = decltype(is_constexpr_impl<p>(0));
constexpr int f() { return 0; }
int g() { return 0; }
static_assert(is_constexpr<f>());
static_assert(!is_constexpr<g>());
Live demonstration.
The key here is that int{(expr, 0U)} contains a narrowing conversion from unsigned int to int (and thus is ill-formed), unless expr is a constant expression, in which case the entire expression (expr, 0U) is a constant expression whose evaluated value fits into the type int.
The following is an implementation of is_constexpr for functions, not for arbitrary expressions, for C++11 and C++17. It requires the arguments to the function you want to test to be default constructible, though.
#include <type_traits>
struct A {}; // don't make it too easy, use a UDT
A f1(A a) { return a; } // is_constexpr -> false
constexpr A f2(A a) { return a; } // is_constexpr -> true
// The following turns anything (in our case a value of A) into an int.
// This is necessary because non-type template arguments must be integral
// (likely to change with C++20).
template <class T> constexpr int make_int(T &&) { return 0; }
// Helper to turn some function type (e.g. int(float)) into a function
// pointer type (e.g. int (*)(float)).
template <class T> struct signature_from;
template <class R, class... Args> struct signature_from<R(Args...)> {
using type = R(*)(Args...);
};
// See std::void_t for the idea. This does it for ints instead of types.
template <int...> using void_from_int = void;
// The fallback case: F is not a function pointer to a constexpr function
template <class T, typename signature_from<T>::type F, class = void_from_int<>>
struct is_constexpr {
static constexpr bool value = false;
};
// If void_from_int<make_int(F(Args()...))> doesn't lead to a substitution
// failure, then this is the preferred specialization. In that case F must
// be a function pointer to a constexpr function. If it is not, it could
// not be used in a template argument.
template <class R, class... Args, typename signature_from<R(Args...)>::type F>
struct is_constexpr<R(Args...), F, void_from_int<make_int(F(Args()...))>>
{
static constexpr bool value = true;
};
// proof that it works:
static_assert(!is_constexpr<A(A), f1>::value, "");
static_assert( is_constexpr<A(A), f2>::value, "");
#if __cplusplus >= 201703
// with C++17 the type of the function can be deduced:
template<auto F> struct is_constexpr2 : is_constexpr<std::remove_pointer_t<decltype(F)>, F> {};
static_assert(!is_constexpr2<f1>::value, "");
static_assert( is_constexpr2<f2>::value, "");
#endif
See it in action at https://godbolt.org/g/rdeQme.
C++20 added std::is_constant_evaluated()
This allows checking if a certain expression is a constant evaluated expression, i.e. being evaluated at compile time.
Usage example:
constexpr int foo(int num) {
// below is true in case the condition is being evaluated at compile time
// side note, using: if constexpr (std::is_constant_evaluated())
// would be evaluated always to true, so you should use a simple if!
if (std::is_constant_evaluated()) {
return foo_compiletime(num);
}
else {
return foo_runtime(num);
}
}
int main() {
constexpr auto t1 = foo(6); // reaches foo_compiletime
const auto t2 = foo(6); // reaches foo_compiletime
int n = rand() % 10;
const auto t3 = foo(n); // reaches foo_runtime
auto t4 = foo(6); // unfortunately, reaches foo_runtime
}
The last call in the example above would reach foo_runtime, since the call is not within a constant expression context (the result is not being used as a constant expression, see also this SO answer).
This may lead to undesired pessimization, compared to the case of leaving the decision to the user, who may call:
auto t4 = foo_compiletime(6);
And the compiler is allowed to perform the operations inside foo_compiletime at compile time, if it is declared as constexpr function, or would be obliged to do that if it is declared consteval. However, once we leave the decision to the compiler, we will reach foo_runtime, unless we explicitly direct the compiler to go for foo_compiletime, by taking the result into a const, constexpr or constinit variable. Which then, in a way, omits the value of having one function for both scenarios, if the user is required to help the compiler peek the right path.
Another possible option for the call to be optimized, is:
constexpr auto temp = foo(6); // foo_compiletime
auto t4 = temp;
But again, we require the user to be aware of the inner behavior of foo, which is not exactly what we want to achieve.
See the pessimization in this code.
See more on that in this great blog post on the subject.