为什么通过显式构造函数提升整数？

Question

这段代码

#include <cstdint>

constexpr uint32_t ticksPerSecond = 100000;

struct timemeasure {
    constexpr explicit timemeasure(uint64_t c) : ticks(c) { }
    uint64_t ticks;
    constexpr timemeasure() : ticks(0) { }
};

struct time : timemeasure {
    static volatile time now;

    constexpr time() : timemeasure() { }
    explicit time(uint64_t c) : timemeasure(c) { }

    // Needed for `x = time::now`
    explicit time(const volatile time &t) : timemeasure(t.ticks) { }
    time& operator=(volatile time t) { ticks = t.ticks; return *this; }
};

inline time foo(const time& t) { return time(t.ticks + 1); }

给出编译错误:

prog.cpp: In function ‘time foo(const time&)’:
prog.cpp:22:57: error: no matching function for call to ‘time::time(time)’
prog.cpp:22:57: note: candidate is:
prog.cpp:14:15: note: constexpr time::time()
prog.cpp:14:15: note:   candidate expects 0 arguments, 1 provided

为什么time::time(time)要打电话？当然uint64_t不允许升职？

Answer 1

复制构造函数time标记为explicit,当从函数返回值时,临时从该值复制初始化.C++ 11标准第8.5/15段规定:

在表单中发生的初始化

T x = a;

以及参数传递,函数返回,抛出异常(15.1),处理异常(15.3)和聚合成员初始化(8.5.1)称为复制初始化.[...]

但是,标记为的构造函数explicit不在复制初始化的上下文中考虑(见13.3.1.4/1和13.3.1.5/1),因此是错误.