boost::enable_if and constructors

[Christoph Bartoschek]

Hi,

I try to use boost::enable_if inside of a constructor but the following code
fails. How can I correct it?


#include <boost/utility/enable_if.hpp>
#include <boost/type_traits/is_base_and_derived.hpp>


template <typename T>
struct Test {
Test() {}

template <typename U>
Test(typename boost::enable_if<

>::type )

{}


};

class Base {}; class Derived : public Base {};

int main() {

Test<Derived> d;
Test<Base> hb(d);
}

Greetings
Christoph Bartoschek

 

[Alf P. Steinbach]

* Christoph Bartoschek:

I try to use boost::enable_if inside of a constructor but the following code
fails. How can I correct it?


#include <boost/utility/enable_if.hpp>
#include <boost/type_traits/is_base_and_derived.hpp>


template <typename T>
struct Test {
Test() {}

template <typename U>
Test(typename boost::enable_if<

{}


};

class Base {}; class Derived : public Base {};

int main() {

Test<Derived> d;
Test<Base> hb(d);
}

First, please see the FAQ item titled "How do I post a question about
code that doesn't work correctly?", currently at <url:
http://www.parashift.com/c++-faq-lite/how-to-post.html#faq-5.8>.

For example, "fails" is not exactly a precise description of the problem.

Having said that, the problem is simply that the compiler can't deduce
the template parameter type. From the compiler's point of view the type
specification is a one-way cryptographic function, sort of, which it's
unable to unscramble. Instead of using enable_if directly as the type
of the formal argument, add a dummy pointer argument using enable_if.

 

[Christoph Bartoschek]

Having said that, the problem is simply that the compiler can't deduce
the template parameter type. From the compiler's point of view the type
specification is a one-way cryptographic function, sort of, which it's
unable to unscramble. Instead of using enable_if directly as the type
of the formal argument, add a dummy pointer argument using enable_if.

Thanks for your answer. I will go the dummy argument way, however I wonder
whether there is a runtime penalty by using such an unused parameter.

Greetings
Christoph Bartoschek

 

[David Harmon]

On Sun, 03 Dec 2006 22:01:32 +0100 in comp.lang.c++, Christoph

Thanks for your answer. I will go the dummy argument way, however I wonder
whether there is a runtime penalty by using such an unused parameter.

A guess: probably zero penalty since Test::Test() is inline.