May 2016

Quick Q: was raw-pointer constructor of shared_ptr a mistake?

Quick A: No, its usage is well defined.

Recently on SO:

was raw-pointer constructor of shared_ptr a mistake?

In hindsight, given make_shared, would shared_ptr have a constructor that takes a raw pointer had it been introduced with C++11?
What if you don't control the allocation of the object? What if you need to use a custom deleter? What if you need list-initialization instead of parens?

None of these cases is handled by make_shared.

Additionally, if you're using weak_ptr, a shared_ptr allocated via make_shared won't free any memory until all the weak_ptrs are destroyed as well. So even if you have a normal shared pointer where none of the above apply, it's possible that you may still prefer the raw pointer constructor.

Yet another situation would be if your type provides overloads for operator new and operator delete. These may make it ill-suited for make_shared, since those overloads will not be called - and presumably they exist for a reason.

CppCon 2015 C++ Multi-dimensional Arrays…—Pramod Gupta

Have you registered for CppCon 2016 in September? Don’t delay – Early Bird registration is open now.

While we wait for this year’s event, we’re featuring videos of some of the 100+ talks from CppCon 2015 for you to enjoy. Here is today’s feature:

C++ Multi-dimensional Arrays...

by Pramod Gupta

(watch on YouTube) (watch on Channel 9)

Summary of the talk:

The language feature of passing a multi-dimensional array to a function without specifying all its dimensions at compile time is crucial for computational physics and applied mathematics. For example a matrix is a two dimensional array and a matrix inversion function which needs to know the size of the matrix at compile time would be of limited use. Major general purpose languages such as C, Java and C# support this feature. Of course, scientific programming languages like Fortran, Matlab and R also support this feature.

C++ is perhaps the only major programming language which does not allow passing a multi-dimensional array to a function unless the size of all the dimensions except the first one is known at compile time. Due to this limitation of C++, various libraries have been developed for using multi-dimensional arrays in C++. Some of these libraries are Blitz++, Armadillo, Eigen and boost.multi_array. These libraries are very large and complex. While they do provide a wide variety of features, they have a learning curve which may be difficult to justify for something as basic as passing multi-dimensional arrays to functions. Also the computational physics or applied mathematics code becomes dependent on a large non-standard library. Hence its usage will be limited to only those scientists who are willing to install these non-standard libraries.

The reference feature of C++ allows us to develop a multi-dimensional array class. The class has a small number of lines of code and hence the code can be included with the scientific application code. We use this class to write programs for various areas of computational physics and show that the class is easy to use and it leads to readable programs.

Quick Q: What makes moving objects faster than copying?

Quick A: it allows to steal ressources instead of creating new ones.

Recently on SO:

What makes moving objects faster than copying?

It's all about implementation. Consider simple string class:

class my_string {
  char* ptr;
  size_t capacity;
  size_t length;
};

Semantics of copy requires us to make a full copy of string including allocation of another array in dynamic memory and copying *ptr contents there, which is expensive.

Semantics of move requires us only to transfer the value of pointer itself to new object without duplicating contents of string.

If, of course, class doesn't use dynamic memory or system resources, then there is no difference between moving and copying in terms of performance.

Quick Q: Why assigning a value is calling the implicit constructor?

Quick A: Because the default assignement operator knows only the type of your object, which it can get be calling the constructor.

Recently on SO:

What is the point of calling constructor with implicit conversion instead of assignment operator after an object is initalized?

There is no implicit CBox::operator=(double), so box = 2.0; has to create a temporary CBox object. It's equivalent to box = CBox(2.0);.

Making your constructor explicit disallows the implicit conversion from double to CBox, so no appropriate assignment operator exists, and you get a compile error.

Quick Q: Why is there no to_string(const string&)?

Quick A: If you need such a function you can create it.

Recently on SO:

Why is there no to_string(const string&)?

You can just write your own templated function with proper overloads as follows:

#include <iostream>
#include <string>
using namespace std;

template<typename T>
std::string toString(const T& t) {
    return std::to_string(t);
}

std::string toString(const char* t) {
    return t;
}

std::string toString(const std::string& t) {
    return t;
}

int main() {
    cout << toString(10) << endl;
    cout << toString(1.5) << endl;
    cout << toString("char*") << endl;
    cout << toString(std::string("string")) << endl;
    return 0;
}

Quick Q: Being smart with smart pointers: avoiding shared_ptr overuse

Quick A: Use unique_ptr when you can.

Recently on SO:

Being smart with smart pointers: avoiding shared_ptr overuse

To allow unique_ptr/shared_ptr, you may use template:

// Dispatcher for make_unique/make_shared
template <template <typename...> class Ptr, typename T>
struct make_helper;

template <typename T>
struct make_helper<std::unique_ptr, T>
{
    template <typename ...Ts>
    std::unique_ptr<T> operator() (Ts&&... args) const {
        return std::make_unique<T>(std::forward<Ts>(args)...);
    }
};

template <typename T>
struct make_helper<std::shared_ptr, T>
{
    template <typename ...Ts>
    std::shared_ptr<T> operator() (Ts&&... args) const {
        return std::make_shared<T>(std::forward<Ts>(args)...);
    }
};

template <template <typename...> class Ptr, typename T, typename ... Ts>
auto make(Ts&&... args)
{
    return make_helper<Ptr, T>{}(std::forward<Ts>(args)...);
}

And then

bool get_resource_parameters(Param1& param1,..., ParamN& paramN)
{
    //...
}

template <template <typename...> class Ptr>
Ptr<resource> open_resource(...)
{
   Param1 param1;
   ...
   ParamN paramN;
   if(!get_resource_parameters(param1, ..., paramN))
       return nullptr;

   return = make<Ptr, resource>(param1, ..., paramN);
}

And check for nullptr instead of split bool and smart_pointer.

RAII versus Exceptions—Arne Mertz

There is no fight, is there?

RAII versus Exceptions

by Arne Mertz

From the article:

Recently I received a question on Twitter whether to prefer RAII over Exceptions. I have seen similar questions being asked again and again over time, so there seems to be some need for clarification...

CppCon 2015 C++ Metaprogramming: A Paradigm Shift—Louis Dionne

Have you registered for CppCon 2016 in September? Don’t delay – Early Bird registration is open now.

While we wait for this year’s event, we’re featuring videos of some of the 100+ talks from CppCon 2015 for you to enjoy. Here is today’s feature:

C++ Metaprogramming: A Paradigm Shift

by Louis Dionne

(watch on YouTube) (watch on Channel 9)

Summary of the talk:

Most people think metaprogramming is hard. It isn't; we just didn't have the right tools for it. This talk will present a new way of metaprogramming using the same syntax as that of normal C++. It will show how the runtime and the compile-time boundaries can be crossed almost seamlessly. It will show how compilation times can be reduced without sacrificing expressiveness. It will introduce Hana [1], a newly accepted Boost library using cutting edge features of the language in a creative way to solve the problem of metaprogramming for good.