减轻头文件中的长限制 [英] Alleviate long qualifications in header files

问题描述

我在头文件中定义了一个 inline 函数对象，如下所示:

I define an inline function object in a header file, like this:

// fmap.hpp
namespace util {
    inline auto constexpr fmap = boost::hana::curry<2>(boost::hana::flip(boost::hana::transform));
}

客户端代码可以简单地#include "fmap.hpp" 并开始使用 util::fmap.

Client code can simply #include "fmap.hpp" and start using util::fmap as they like.

到目前为止一切顺利.

但有时这些对象的定义读起来很麻烦，如果它们充满了qui::quo::qua::lify.

But sometimes the definition of such objects can be cumbersome to read, if they are so full of qui::quo::qua::lify.

我该如何缓解这种情况?

How can I alleviate this?

理想情况下，我希望 fmap 的定义如下所示:

Ideally, I'd like the definition of fmap to look like this:

namespace util {
    inline auto constexpr fmap = curry<2>(flip(transform));
}

但同时我不想把 using namespace boost::hana; 放在顶层，因为客户端代码的命名空间会被 boost::hana(或来自其他命名空间)名称.

but at the same time I don't want to put a using namespace boost::hana; at top level, as client code's namespace would be cluttered with boost::hana (or from other namespaces) names.

在这种情况下，是否有一些 C++ 指南或最佳实践可以派上用场?

Is there some C++ guideline or best practice that comes handy in this situation?

推荐答案

我的其他自我回答很好，但它也创建了一个虚假的用户可见类 Fmap，我真的不需要存在.

My other self answer is good, but it too creates a spurious and user-visible class Fmap which I really don't need to exist.

考虑一下，我可以通过构建并动态调用从 void 到所需 lambda 的 lambda 来解决这个谜语:

Thinking about it, I can solve the riddle by just constructing and calling on the fly a lambda from void to the desired lambda:

namespace utils {

inline auto constexpr fmap = [](){
    using namespace boost::hana;
    return curry<2>(flip(transform));
}();

}

这篇关于减轻头文件中的长限制的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！