问题描述
可以通过访问它的 operator()
来推断非泛型 lambda 的 arity.
It is possible to deduce arity of a non-generic lambda by accessing its operator()
.
template <typename F>
struct fInfo : fInfo<decltype(&F::operator())> { };
template <typename F, typename Ret, typename... Args>
struct fInfo<Ret(F::*)(Args...)const> { static const int arity = sizeof...(Args); };
这对于 [](int x){ return x;}
因为 operator()
不是模板化的.
This is nice and dandy for something like [](int x){ return x; }
as the operator()
is not templated.
但是,泛型 lambda 对 operator()
进行模板化,并且只能访问模板的具体实例 - 这有点问题,因为我无法手动为 operator()
提供模板参数code>operator() 因为我不知道它的 arity 是什么.
However, generic lambdas do template the operator()
and it is only possible to access a concrete instantiation of the template - which is slightly problematic because I can't manually provide template arguments for the operator()
as I don't know what its arity is.
所以,当然,像
auto lambda = [](auto x){ return x; };
auto arity = fInfo<decltype(lambda)>::arity;
不起作用.
我不知道要转换为什么,也不知道要提供哪些模板参数(或提供多少)(operator()?>
).
任何想法如何做到这一点?
I don't know what to cast to nor do I know what template arguments to provide (or how many) (operator()<??>
).
Any ideas how to do this?
推荐答案
这是不可能的,因为函数调用运算符可以是可变参数模板.一般来说,对于函数对象永远不可能这样做,特殊情况下的 lambdas 因为它们碰巧不是同样强大,所以总是一个坏主意.现在是时候让这个坏主意回家了.
It's impossible, as the function call operator can be a variadic template. It's been impossible to do this forever for function objects in general, and special-casing lambdas because they happened to not be equally powerful was always going to be a bad idea. Now it's just time for that bad idea to come home to roost.
这篇关于泛型 lambda 的数量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跟版网!