Tag: c++17

C++ multiple template parameter packs

The idea

This is just a short idea for multiple template parameter packs on a specific use case. While there are other more generic ways to achieve this, I found a method that is easier to digest for my case.

One of my learning projects is a map-like container with infinite depth, multiple specific types of keys, and any type of value.

The need for multiple template parameter packs came when I wanted to be more specific about “any type of value”. “Any” is… any. Nothing specific, clear, or well-known. And I wanted more clarity.

My map is declared as:

msd::poly_map<int, double, std::string> map;

The template arguments are the types of keys. No types for the values because the map can hold any value. But I want to be as specific as I am for the keys. What I need is to separate the key types from the value types. I want two sets of template parameters. How could I tell them apart?

The solution

After a few minutes of diving in, the idea that popped up is to store the values exactly how I store the keys: inside a variant. The bonus for switching from any to variant is that:

Variant is not allowed to allocate additional (dynamic) memory.

I introduced an auxiliary type to represent a multiple-parameter pack. And I passed two of these to my map: one for keys, one for values.

template<typename... Types>
struct types {
    using types_ = std::variant<Types...>;
};

template<typename Keys, typename Values>
struct poly_map;

poly_map<types<int, double>, types<int, bool>> map;

The full source code

Everything put together in a raw version looks like:

#include <cassert>
#include <map>
#include <variant>

template<typename... Types>
struct types {
    using types_ = std::variant<Types...>;
};

template<typename... Types>
using keys = types<Types...>;

template<typename... Types>
using values = types<Types...>;

template<typename Keys, typename Values>
struct poly_map {
    std::map<typename Keys::types_, poly_map> items_;

    using value_types = typename Values::types_;
    value_types value_;

    template <typename T>
    auto& operator=(T&& v)
    {
        static_assert(std::is_constructible_v<value_types, T>, "wrong value type");

        value_ = std::forward<T>(v);

        return *this;
    }

    template <typename T>
    auto& operator[](const T key)
    {
        return items_[key];
    }

    template <typename T>
    auto& get() const
    {
        static_assert(std::is_constructible_v<value_types, T>, "wrong value type");

        return std::get<T>(value_);
    }
};

struct X {
    int v{};
};

int main() {
    poly_map<keys<int, double>, values<int, bool, X>> map;

    map[1] = true;
    assert(map[1].get<bool>());

    map[2] = X{1};
    assert(map[2].get<X>().v == 1);
    
    //map[1] = 0.1; // does not compile because map can't hold double as value

    map[1][2.2] = 14;
    assert(map[1][2.2].get<int>() == 14);
}

Calling multiple functions for an input

Simply put, when I opt for a data-driven design, I separate the data from the behavior. Given an input as

struct Input {
    int value;
};

I pass it to some components that operate on it. I… call some functions.

void set(Input& in, int value) { input.value = value; }
void reset(Input& in) { input.value = 0; }

Input in{};
set(in, 2);
reset(in);

 

Because life is better with patterns, I’d want to have independent and configurable functions and a clear intent of their role and usage. Short story, a way to do this is a list of functions to be called with an input.

template<typename T, typename... Fs>
void apply(T& in, Fs&&... fs) {
    (fs(in), ...);
}

I’ve used the C++17 fold expression to unpack the template parameters (the list of functions). Continue reading Calling multiple functions for an input

Compile-time recursion in C++17

While not a large upgrade to the C++ standard, C++17 brought some important features. One of them helps me have simpler code in the context of compile-time recursion.

I’ve experimented with some strategies to iterate over a tuple. I wanted a list of different types that I could iterate over as I would do with an array. Without using dynamically allocated memory. Some kind of static polymorphism.

The previous article on this topic gives an implementation for C++14. Which is not that complicated, but whenever I can get something simpler, I’m very interested. That article is the base for this one, thus reading it before will give more context.

I want to improve the iteration that is based on compile-time recursion. More specifically, the exit case of the recursion. The C++14 implementation is based on template specialization. There are two versions of the iterator struct: one iterates over the elements of the tuple except for the last one, and the second one is for the last element, where the iteration must stop.

template<typename T, std::size_t S = std::tuple_size<T>::value, std::size_t I = S - 1>
struct Iterator {
    template<typename C>
    void operator()(T& objects, C callback) {
        callback(std::get<S - I - 1>(objects));
        Iterator<T, S, I - 1>{}(objects, callback);
    }
};

template<typename T, std::size_t S>
struct Iterator<T, S, 0> {
    template<typename C>
    void operator()(T& objects, C callback) {
        callback(std::get<S - 1>(objects));
    }
};

The code is partially duplicated and not the easiest to understand. Any piece of code that can be deleted is the best code I can get. And constexpr if lets me do just that. I can delete one of the structs and have the implementation, including the exit case, in one struct. The constexpr if feature allows the use of an if statement in more complex compile-time cases.

template<typename T, std::size_t I = 0U>
struct Iterator {
    template<typename C>
    void operator()(T& objects, C callback) {
        if constexpr (I < std::tuple_size_v<T>) {
            callback(std::get<I>(objects));
            Iterator<T, I + 1U>{}(objects, callback);
        }           
    }
};

I have an easier-to-read code. The iteration starts naturally from zero, and as long as I’m not past the last element, I apply the given callback, then get to the next element.

Simplicity is always welcome.

A bitfield abstraction for enum values

Given different values represented by an enum, the requirement is to pass a list of those values to another system. It’s not mandatory to use an enum, I’ve chosen it just as a use case. The list of values can be a vector, an array, a bitset, or any other container or utility that can hold multiple values. Depending on the context, there are advantages and disadvantages over one container or another.

Options

A vector is very easy to use but uses dynamic memory allocation. In an embedded context, the use of dynamic memory can be restricted.

An array uses preallocated memory and has a fixed number of elements, which could be the maximum number of values in the enum. If you want to send fewer elements than the maximum, you must have a convention to let the other system know how many you are sending. This is because you will always send a fixed number of elements (the maximum). You can choose a special value that indicates an element in the array is not of interest. Or you can place the values you want to send starting with the first position in the array, and pass along another value that says how many elements you are sending.

For a bitset, you must know the number of bits used. And you have a general semantic of manipulating values. If these aspects are convenient, a bitset can be an option.

The most simple option I can think of is a bitwise representation on an unsigned, fixed-width integer type (eg: uint32_t). It can give you the smallest memory space to represent multiple values. If you need this list of values in a very small scope, like a small function where you set the bits on a variable which you pass to the other system, it might be enough. If you pass this variable in larger scopes of the project, it’s a matter of time until someone does not know what that variable holds (there is no semantic). And they might confuse it for a variable that holds a single value, not a representation of multiple ones. Then, operations like equality might work by coincidence in some cases, other cases being runtime bugs.

For all of the above and other options not mentioned, you need to obtain the underlying value of the enum. An unscoped enum is implicitly converted to the numeric type you use for the option you choose. From a scoped enum you have to explicitly get the value. Continue reading A bitfield abstraction for enum values