我一直在看max method在 Ruby 的 Enumerable混合(v2.4.1)。

这是一个相当简单的方法，但是当存在重复项时它如何排序项目有点令人困惑。

例如:

x = [1,2,3,4,5,6,7,8,9]
x.max {|a, b| a%2 <=> b%2}
=> 1
10.times{|y| p x.max(y) {|a, b| a%2 <=> b%2}}
[]
[1]
[1, 7] # why is 7 the next element after 1?
[3, 1, 5] # why no more 7?
[7, 3, 1, 5] # 7 is now first
[9, 7, 3, 1, 5]
[9, 7, 3, 1, 5, 6]
[9, 7, 3, 1, 5, 4, 6]
[9, 7, 3, 1, 5, 2, 4, 6]
[9, 7, 5, 3, 1, 8, 6, 4, 2] # order has changed again (now seems more "natural")

最佳答案

您可以通过使用 max_by 来简化您的示例以产生类似的结果:
10.times{|y| p x.max_by(y) {|t| t%2}}
我花了一些时间在源上，但找不到任何漏洞。

在我记得看到一个名为 Switch: A Deep Embedding of Queries into Ruby 的出版物之后(Manuel Mayr 的论文)我找到了答案。

您可以在第 104 页的何处找到 max_by 的答案:

... Here, the value in the input list that assumes the maximum when evaluated by the function is returned. If more than one value yields the maximum, the choice for a result among these values is arbitrary. ...

The result is not guaranteed to be stable. When two keys are equal, the order of the corresponding elements is unpredictable.

+Tue Sep 15 12:44:32 2015  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+   * util.c (ruby_qsort): use BSD-style qsort_r if available.

Thu May 12 00:18:19 2016  NAKAMURA Usaku  <usa@ruby-lang.org>

    * win32/Makefile.sub (HAVE_QSORT_S): use qsort_s only for Visual Studio
      2012 or later, because VS2010 seems to causes a SEGV in
test/ruby/test_enum.rb.

#ifdef HAVE_GNU_QSORT_R
#define ruby_qsort qsort_r
#else    void ruby_qsort(void *, const size_t, const size_t,
    int (*)(const void *, const void *, void *), void *);
#endif

    /*
    * To save stack space we sort the smaller side of the partition first
    * using recursion and eliminate tail recursion for the larger side.
    */

     #if defined HAVE_BSD_QSORT_R
    typedef int (cmpfunc_t)(const void*, const void*, void*);

    struct bsd_qsort_r_args {
        cmpfunc_t *cmp;
        void *arg;
    };

    static int
    cmp_bsd_qsort(void *d, const void *a, const void *b)
    {
        const struct bsd_qsort_r_args *args = d;
        return (*args->cmp)(a, b, args->arg);
    }

    void
    ruby_qsort(void* base, const size_t nel, const size_t size, cmpfunc_t *cmp, void *d)
    {
        struct bsd_qsort_r_args args;
        args.cmp = cmp;
        args.arg = d;
        qsort_r(base, nel, size, &args, cmp_bsd_qsort);
    }

x=[1,2,3,4,5,6,7,8,9]
10.times{|y| p x.max_by(y) {|t| t%2}}

The result:
[]
[5]
[7, 1]
[3, 1, 5]
[7, 3, 1, 5]
[9, 7, 3, 1, 5]
[5, 9, 1, 3, 7, 6]
[5, 1, 9, 3, 7, 6, 4]
[5, 1, 3, 7, 9, 6, 4, 2]
[9, 1, 7, 3, 5, 4, 6, 8, 2]

The result:
[]
[1]
[7, 1]
[5, 3, 1]
[5, 7, 3, 1]
[5, 9, 7, 3, 1]
[5, 1, 9, 7, 3, 6]
[5, 1, 3, 9, 7, 4, 6]
[5, 1, 3, 7, 9, 2, 6, 4]
[9, 1, 3, 7, 5, 8, 4, 6, 2]

The result:
[]
[1]
[9, 7]
[1, 7, 3]
[5, 1, 7, 3]
[5, 1, 3, 9, 7]
[7, 5, 9, 3, 1, 2]
[7, 9, 5, 3, 1, 2, 4]
[7, 9, 3, 1, 5, 2, 4, 8]
[5, 9, 1, 3, 7, 4, 6, 8, 2]

The result:
[]
[9]
[9, 7]
[3, 1, 7]
[3, 5, 1, 7]
[7, 5, 1, 3, 9]
[7, 9, 5, 1, 3, 2]
[7, 9, 3, 5, 1, 4, 2]
[7, 9, 3, 1, 5, 8, 2, 4]
[5, 9, 3, 1, 7, 2, 4, 6, 8]

The result:
[]
[3]
[1, 1]
[9, 1, 7]
[3, 9, 1, 7]
[5, 3, 9, 1, 7]
[1, 5, 3, 9, 1, 7]
[5, 9, 3, 7, 1, 1, 1]
[1, 5, 9, 1, 7, 1, 3, 4]
[1, 1, 5, 9, 1, 7, 3, 4, 2]
[1, 1, 1, 5, 7, 3, 9, 4, 2, 8]
[9, 1, 7, 1, 5, 3, 1, 2, 6, 8, 4]

The Result:
[]
[1]
[1, 1]
[7, 9, 1]
[7, 3, 9, 1]
[7, 5, 3, 9, 1]
[7, 1, 5, 3, 9, 1]
[1, 5, 9, 3, 7, 1, 1]
[1, 1, 5, 9, 3, 7, 1, 4]
[1, 1, 1, 5, 9, 3, 7, 2, 4]
[1, 7, 3, 1, 5, 9, 1, 2, 4, 8]
[9, 3, 1, 7, 1, 5, 1, 2, 8, 6, 4]

Warning: If two objects compare as equal, their order after sorting is unpredictable. That is to say, the sorting is not stable. This can make a difference when the comparison considers only part of the elements. Two elements with the same sort key may differ in other respects.

关于c - Ruby 的最大函数顺序如何重复？，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/46286715/