开发者

Should std::sort work with lambda function in c++0x/c++11?

开发者 https://www.devze.com 2023-04-12 20:24 出处:网络
I tried to use lambda function with sort, but was getting \"Segmentation fault开发者_JAVA技巧\" errors. I managed to simplify the code to the following:

I tried to use lambda function with sort, but was getting "Segmentation fault开发者_JAVA技巧" errors. I managed to simplify the code to the following:

#include <iostream>
#include <algorithm>

int main()
{
  const int len = 18;
  int intArr[len];
  for (int i=0;i<len;i++) intArr[i]=1000+i;
  // The following is expected to sort all but the last element of the array
  std::sort(intArr, intArr + len -1, [](int a, int b)
    {
      std::cout<<"("<<a<<", "<<b<<")\n";
      return (a<b?-1:(a>b?1:0));
    });
  return 0;
}

I compile and run this code in Ubuntu 11.04 (x64) using

g++ -std=gnu++0x test2.cpp && ./a.out.

It prints a lot of pairs of the form (large_integer, 1008), a couple of (0, 1008) and exits with "Segmentation fault".


The comparison predicate should return a bool: true if a < b and false otherwise. Change the return statement to:

  return a < b;

Not to be confused with C-style 3-way comparison functions.

NOTE: Though C++20 does introduce a 3-way comparison operator <=>, sort would still expect a 2-way compare predicate.


The predicate is supposed to implement a simple, weak ordering. Also your range is off if you want to sort the entire thing. (I missed that that was intentional.) So all in all we're looking for something like this:

std::sort(intArr, intArr + nelems, [](int a, int b){ return a < b; });

Or even:

std::sort(intArr, intArr + nelems);

The default predicate for sorting is std::less<T>, which does exactly what the lambda does.


The predicate for std::sort doesn't take the Java-like -1,0,1, but instead wants you to return a boolean that answers the question 'Is the first argument less than the second argument?', which is used to weakly order the elements. Since -1 is a non-zero value, it is considered true by the sort algorithm, and it causes the algorithm to have a breakdown.

0

精彩评论

暂无评论...
验证码 换一张
取 消

关注公众号