r****t
发帖数: 10904 | 1 这个应该是常见问题了，google 了一下确没有找到，这段小程序不能 compile:
template
class A {
public:
A(float numbers[N]):
numbers(numbers) {}
float numbers[N]; // this does not compile, see error msg below.
//float* numbers; // this works fine.
};
float arr[3] = {1.,2.,3.};
int main() {
A<3> a(arr);
}
compile 错误是：
test3.cpp: In constructor 'A::A(float*) [with int N = 3]':
test3.cpp:14: instantiated from here
test3.cpp:5: error: incompatible types in assignment | x****u
发帖数: 44466 | 2 数组不是C++的类型，放在参数里面就等于const pointer，你用union包装一下吧。
below.
【在 r****t 的大作中提到】
: 这个应该是常见问题了，google 了一下确没有找到，这段小程序不能 compile:
: template
: class A {
: public:
: A(float numbers[N]):
: numbers(numbers) {}
: float numbers[N]; // this does not compile, see error msg below.
: //float* numbers; // this works fine.
: };
: float arr[3] = {1.,2.,3.};
| r****t
发帖数: 10904 | 3 用union包装是啥意思呢？展开说说？
【在 x****u 的大作中提到】
: 数组不是C++的类型，放在参数里面就等于const pointer，你用union包装一下吧。
:
: below.
| p***o
发帖数: 1252 | 4 memcpy or std::copy
below.
【在 r****t 的大作中提到】
: 这个应该是常见问题了，google 了一下确没有找到，这段小程序不能 compile:
: template
: class A {
: public:
: A(float numbers[N]):
: numbers(numbers) {}
: float numbers[N]; // this does not compile, see error msg below.
: //float* numbers; // this works fine.
: };
: float arr[3] = {1.,2.,3.};
| r****t
发帖数: 10904 | 5 我觉得在 ctor 里面 memcpy 应该是个好办法。
【在 p***o 的大作中提到】
: memcpy or std::copy
:
: below.
| t****t
发帖数: 6806 | 6 first of all, you can not copy array by using numbers(numbers). must use
memcpy or std::copy.
second, if you declare ctor as A::A(float numbers[N]), it is equivalent to A
float[1] without compiling errors, which may not be desirable. depending on
your requirement, you may want to use
A::A(const float (&numbers)[N])
as ctor, which preserves length information and enforces that you can only
send in float[N] at compile time.
below. | r****t
发帖数: 10904 | 7
A
on
对，这个是唯一用一个 argument 还能 enforce N 的办法
【在 t****t 的大作中提到】
: first of all, you can not copy array by using numbers(numbers). must use
: memcpy or std::copy.
: second, if you declare ctor as A::A(float numbers[N]), it is equivalent to A
: float[1] without compiling errors, which may not be desirable. depending on
: your requirement, you may want to use
: A::A(const float (&numbers)[N])
: as ctor, which preserves length information and enforces that you can only
: send in float[N] at compile time.
: below.
|
|
