boost multi_index 提取的键是否被缓存?

发布于 2024-10-21 14:45:02 字数 266 浏览 4 评论 0 原文

我正在使用 boost::multi_index 以及我想根据其大小进行索引的数据类型。然而,该数据类型的 size() 成员函数的执行成本很高。 multi_index 是否缓存从其键提取器获取的值?

例如,如果我创建了一个带有有序索引和成员函数键 (element.size()) 的 multi_index 容器,并插入了一个元素,该元素的大小将其放置在容器中间的某个位置,那么容器是否会重新调用 size () 成员函数对其访问的所有元素进行遍历,同时遍历其内部数据结构,然后找到正确的插入点?

I am using boost::multi_index with a data type I'd like to index based on its size. However, the size() member function of this data type is expensive to execute. Does multi_index cache the values it gets from its key extractors?

For example, if I created a multi_index container with an ordered index with a member function key (element.size()), and inserted an element whose size put it somewhere in the middle of the container, would the container re-invoke the size() member function on all the elements it visits while traversing its internal data structure before finding the right insertion point?

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(1

澜川若宁 2024-10-28 14:45:02

好吧,成员函数索引器的文档说它们调用引用的成员函数: http://www.boost.org/doc/libs/1_46_0/libs/multi_index/doc/reference/key_extraction.html#key_extractors

但是当有疑问时,配置文件:

#include <boost/multi_index_container.hpp>
#include <boost/multi_index/mem_fun.hpp>
#include <boost/multi_index/indexed_by.hpp>
#include <boost/multi_index/hashed_index.hpp>
#include <boost/multi_index/ordered_index.hpp>
#include <iostream>
#include <vector>
#include <algorithm>

using namespace std;

namespace bmi = boost::multi_index;

int g_calls = 0;
struct A
{
  explicit A(int sz) : m_size(sz) { }
  int size() const { ++g_calls; return m_size; }
private:
  int m_size;
};

typedef boost::multi_index_container<
  A*,
  bmi::indexed_by<
    bmi::ordered_non_unique<
      BOOST_MULTI_INDEX_CONST_MEM_FUN(A,int,A::size)
    >
  >
> container_t;

int main()
{
  container_t cont;
  int n = 100;
  vector<int> o(2*n+1);
  for( int i = 0; i != 2*n+1; ++i )
    o[i] = i;
  random_shuffle(o.begin(), o.end());

  for( int i = 0; i != n; ++i )
    cont.insert(new A(o[i]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  for( int i = n+1; i <= 2*n; ++i )
    cont.insert(new A(o[i]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  cont.insert(new A(o[n]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  for( int i = 0; i != o.size(); ++i )
    cont.find(o[i]);
  cout << "Calls after calling find " << o.size() << " times: "<< g_calls << endl;
  return 0;
}

给出以下内容输出(使用 Boost 1.46):

Calls to A::size(): 629
Calls to A::size(): 1465
Calls to A::size(): 1474
Calls after calling find 201 times: 3262

所以,答案似乎是“不,它不缓存值”。

Well, the documentation for member function indexers says they call the referenced member function: http://www.boost.org/doc/libs/1_46_0/libs/multi_index/doc/reference/key_extraction.html#key_extractors

But when in doubt, profile:

#include <boost/multi_index_container.hpp>
#include <boost/multi_index/mem_fun.hpp>
#include <boost/multi_index/indexed_by.hpp>
#include <boost/multi_index/hashed_index.hpp>
#include <boost/multi_index/ordered_index.hpp>
#include <iostream>
#include <vector>
#include <algorithm>

using namespace std;

namespace bmi = boost::multi_index;

int g_calls = 0;
struct A
{
  explicit A(int sz) : m_size(sz) { }
  int size() const { ++g_calls; return m_size; }
private:
  int m_size;
};

typedef boost::multi_index_container<
  A*,
  bmi::indexed_by<
    bmi::ordered_non_unique<
      BOOST_MULTI_INDEX_CONST_MEM_FUN(A,int,A::size)
    >
  >
> container_t;

int main()
{
  container_t cont;
  int n = 100;
  vector<int> o(2*n+1);
  for( int i = 0; i != 2*n+1; ++i )
    o[i] = i;
  random_shuffle(o.begin(), o.end());

  for( int i = 0; i != n; ++i )
    cont.insert(new A(o[i]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  for( int i = n+1; i <= 2*n; ++i )
    cont.insert(new A(o[i]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  cont.insert(new A(o[n]));
  cout << "Calls to A::size(): "<< g_calls << endl;
  for( int i = 0; i != o.size(); ++i )
    cont.find(o[i]);
  cout << "Calls after calling find " << o.size() << " times: "<< g_calls << endl;
  return 0;
}

Gives the following output (using Boost 1.46):

Calls to A::size(): 629
Calls to A::size(): 1465
Calls to A::size(): 1474
Calls after calling find 201 times: 3262

So, it appears the answer is no, it doesn't cache values.

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文