公众号
qq群 点击进入
RSS https://github.com/wanghenshui/cppweeklynews/releases.atom
欢迎投稿,推荐或自荐文章/软件/资源等
请提交 issue 或者评论区留言
另外公众号挂了c++templates 第二版优惠
从上面的链接里下单的兄弟买书到货后可以找我退佣金,加我微信,公众号后台回复即可
本期文章由 黄亮 不语 赞助
标准委员会动态/ide/编译器信息放在这里
编译器信息最新动态推荐关注hellogcc公众号 本周更新 2023-12-20 第233期
委员会邮件 https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2023/#mailing2023-12
本月委员会邮件没有什么新鲜的,顶多fiber_context。这里不展开了
template<auto N> consteval auto nth(auto... ts) { return ts...[N]; }
static_assert(1 == nth<0>(1, 2, 3));
static_assert(2 == nth<1>(1, 2, 3));
static_assert(3 == nth<2>(1, 2, 3));
还不确定有什么作用
const的副作用
经典例子
#include <iostream>
int main() {
const int i = 9;
int& j = const_cast<int&>(i);
j = 4;
std::cout << i << std::endl; // prints 9
std::cout << j << std::endl; // prints 4
}
https://godbolt.org/z/vGG3cdavE
但是这个例子,返回优化了,即使没有实现move
#include <iostream>
#include <cassert>
class Rng {
int _min;
int _max;
// invariant: _min <= _max
public:
Rng(int lo, int hi)
// precond: lo <= hi
: _min(lo), _max(hi)
{ assert(_min <= _max); }
int const& min() const { return _min; }
int const& max() const { return _max; }
void set(int lo, int hi)
// precond: lo <= hi
{
_min = lo;
_max = hi;
assert(_min <= _max);
}
Rng(Rng&&) { assert(false); } // this is never called
Rng(Rng const&) { assert(false); } // this is never called
};
const Rng foo() {
const Rng r {1, 2};
std::cout << &r << std::endl;
return r;
}
Rng bar() {
Rng r = foo();
r.set(3, 4);
std::cout << &r << std::endl;
return r;
}
int main() {
const Rng z = bar();
std::cout << &z << std::endl;
}
https://godbolt.org/z/n9nn5GjMM
注意这两个例子的区别,统一作用域上的修改
上面的这个xyz 本质上就是一个对象,和第一个例子同一个域里const_cast导致变化不同
怎么mock时间?比如特化?
template <typename ...>
constexpr auto clock_impl = std::chrono::some_clock{};
template <typename ... Ts>
struct app_clock {
static
std::chrono::some_clock::time_point now()
{
return clock_impl<Ts...>.now();
}
};
struct test_clock {
using time_point = std::chrono::some_clock::time_point;
static time_point now() { return {};}
};
template <>
constexpr auto clock_impl<> = test_clock{};
https://godbolt.org/z/GbWYaGc7q
讲的不错
写的很有深度,值得一看
省流 arm O3 优化bug
-ftrivial-auto-var-init=[pattern|zero|uninitialized]
帮助自动初始化栈上的局部变量
开销很大,研究了一圈放弃了
一种是make_unique这种需要指定T的,一种是swap sort这种不指定T的
如何跨过这种边界,有设计,比如CTAD,但这并不建议使用
那就只能多提供重载了,比如optional
template<class T, class A>
optional<T> make_optional(A);
template<class A>
optional<A> make_optional(A);
然后她举了个例子,怎么设计强制制定T和忽略T
https://godbolt.org/z/h38PhG3Y6
#include <type_traits>
#include <iostream>
//template<class T, class A>
//T implicitly_convert_to(std::type_identity_t<A>) = delete;
template<class T, class A,
std::enable_if_t<std::is_convertible_v<A, T>, int> E = 0>
T implicitly_convert_to(A arg) { return T(arg); }
int main() {
//auto i0 = implicitly_convert_to(9.9999999);
//std::cout << i0 << "\n";
auto i1 = implicitly_convert_to<int>(9.9999999);
std::cout << i1 << "\n";
//auto j2 = implicitly_convert_to<int, float>(9.9999999);
//std::cout << j2 <<"\n";
return 0;
}
看一乐
直接贴代码了
https://godbolt.org/z/MvYfbEP8r
https://godbolt.org/z/57zsK9rEn
设计的挺有意思的,鉴于篇幅,放在后面
constexpr的代码 编译器没有做充分的优化。这可能加剧编译时长
算是个坑爹细节。运行时能充分优化的代码到了constexpr期反而没优化了
寻路算法,A*之类的,如何缓存友好。STL不太行
valgrind 也可以测试cache性能,判断miss
valgrind --tool=cachegrind --cache-sim=yes
perf也可以,就不说了
结论就是 顺序访问 不要跳转 只访问用到的数据 s执行路径里没有malloc
比如std::unordered_multimap::equal_range 内存不连续,miss就很多
概念很帅,把range推广到分布式,做的一些工作
代码在这里 https://github.com/oneapi-src/distributed-ranges/tree/main
#include <iostream>
#include <vector>
#include <coroutine>
#include <chrono>
#include <thread>
#include <utility>
#include <functional>
#include <memory>
#include <algorithm>
#include <iterator>
#include <atomic>
#define FUNC() std::cout << __func__ << '\n'
namespace details {
template <typename InputIterator>
void printIterable(InputIterator first, InputIterator last) {
using value_type = std::decay_t<decltype(*first)>;
std::cout << '[';
if constexpr (std::is_same_v<std::uint8_t, value_type>) {
std::copy(first, std::prev(last), std::ostream_iterator<std::uint16_t>(std::cout, ", "));
std::cout << static_cast<std::uint16_t>(*std::prev(last)) << "]\n";
} else {
std::copy(first, std::prev(last), std::ostream_iterator<value_type>(std::cout, ", "));
std::cout << *std::prev(last) << "]\n";
}
}
template <typename Container>
void printContainer(const Container& container) {
printIterable(std::cbegin(container), std::cend(container));
}
}
class [[nodiscard]] AudioDataResult final {
public:
class promise_type;
using handle_type = std::coroutine_handle<promise_type>;
// Predefined interface that has to be specify in order to implement
// coroutine's state-machine transitions
class promise_type {
public:
using value_type = std::vector<int>;
AudioDataResult get_return_object() {
return AudioDataResult{handle_type::from_promise(*this)};
}
std::suspend_never initial_suspend() noexcept { return {}; }
std::suspend_always final_suspend() noexcept { return {}; }
void return_void() {}
void unhandled_exception() {
std::rethrow_exception(std::current_exception());
}
// Generates the value and suspend the "producer"
template <typename Data>
requires std::convertible_to<std::decay_t<Data>, value_type>
std::suspend_always yield_value(Data&& value) {
data_ = std::forward<Data>(value);
data_ready_.store(true, std::memory_order_relaxed);
return {};
}
auto await_transform(handle_type other) {
// Awaiter interface: for consumer waiting on data being ready
struct AudioDataAwaiter {
explicit AudioDataAwaiter(promise_type& promise) noexcept: promise_(promise) {}
bool await_ready() const { return promise_.data_ready_.load(std::memory_order_relaxed);}
void await_suspend(handle_type) const {
while(not promise_.data_ready_.exchange(false)) {
std::this_thread::yield();
}
}
value_type&& await_resume() const {
return std::move(promise_.data_);
}
private:
promise_type& promise_;
};//Awaiter interface
return AudioDataAwaiter{other.promise()};
}
private:
value_type data_;
std::atomic<bool> data_ready_;
}; //promise_type interface
explicit operator handle_type() const { return handle_;}
// Make the result type move-only, due to ownership over the handle
AudioDataResult(const AudioDataResult&) = delete;
AudioDataResult& operator=(const AudioDataResult&) = delete;
AudioDataResult(AudioDataResult&& other) noexcept: handle_(std::exchange(other.handle_, nullptr)){}
AudioDataResult& operator=(AudioDataResult&& other) noexcept {
using namespace std;
AudioDataResult tmp = std::move(other);
swap(*this, tmp);
return *this;
}
// d-tor: RAII
~AudioDataResult() { if (handle_) {FUNC(); handle_.destroy();}}
// For resuming the producer - at the point when the data are consumed
void resume() {if (not handle_.done()) { FUNC(); handle_.resume();}}
private:
AudioDataResult(handle_type handle) noexcept : handle_(handle) {}
private:
handle_type handle_;
};
using data_type = std::vector<int>;
AudioDataResult producer(const data_type& data) {
for (std::size_t i = 0; i < 5; ++i) {
FUNC();
co_yield data;
}
co_yield data_type{}; // exit criteria
co_return;
}
AudioDataResult consumer(AudioDataResult& audioDataResult) {
for(;;) {
FUNC();
const auto data = co_await static_cast<AudioDataResult::handle_type>(audioDataResult);
if (data.empty()) {std::cout << "No data - exit!\n"; break;}
std::cout << "Data received:";
details::printContainer(data);
audioDataResult.resume(); // resume producer
}
co_return;
}
int main() {
{
const data_type data = {1, 2, 3, 4};
auto audioDataProducer = producer(data);
std::thread t ([&]{auto audioRecorded = consumer(audioDataProducer);});
t.join();
}
std::cout << "bye-bye!\n";
return 0;
}
最近看 丁胖子金牌讲师视频,非常乐呵,讲师也是学c++ 的
啥也不是,散会!
如果有疑问评论最好在上面链接到评论区里评论,这样方便搜索,微信公众号有点封闭/知乎吞评论