嗯，他们把这些类型区分开来;大概是为了避免这种运行时决策。这至少会导致代码膨胀，因为现在每个算法都必须被示例化。但如果你认为这是必要的，这是我能想到的最短的：

#include <algorithm>
#include <execution>
#include <thread>
#include <variant>

using execution_variant = std::variant<
        std::execution::sequenced_policy, std::execution::parallel_policy>;

execution_variant select_execution(std::size_t items)
{
    if(items > std::thread::hardware_concurrency() * 16)
        return std::execution::par;
    return std::execution::seq;
}

void sort(std::vector<int>& vec)
{
    std::visit([&](auto policy) {
            std::sort(policy, vec.begin(), vec.end()); },
        select_execution(vec.size()));
}

或者用一个助手来使调用网站更简洁：

struct Execution
{
    execution_variant var;

    template<class Fun>
    auto operator()(Fun&& fun) const
    { return std::visit(fun, var); }

    static Execution select(std::size_t items)
    {
        if(items > std::thread::hardware_concurrency() * 16)
            return {std::execution::par};
        return {std::execution::seq};
    }
};
void sort(std::vector<int>& vec)
{
    Execution::select(vec.size())([&](auto policy) {
          std::sort(policy, vec.begin(), vec.end()); });
}