Re: [boost] Anyone downloaded the sorting library from vault?

From: Sam Schetterer (samthecppman_at_[hidden])
Date: 2007-03-16 18:01:53

On March 16, Lewis Hyatt wrote
     I took a look at it. Some of my thoughts:

    -You should provide random access iterator interfaces, not just a bare
pointer.

    -Your docs say you require T to be an integral type or have operator<
defined, I
    think you meant to say "built-in arithmetic type" or something similar.
You
    should also let the user provide their own comparison functor instead of
    requiring them to overload operator<

    -Based on your descriptions, it looks like merge_sort is the same as
    std::stable_sort and quicksort is the same as std::sort, more or less.
(Except
    the standard functions have more generic interfaces). What advantage are
they
    supposed to have over the standard functions? I tried some timings, and
found
    that for both ints and doubles, on my system, your quicksort is about
50% slower
    than std::sort, and your mergesort is about 5% faster than
std::stable_sort, at
    least for the case of already-sorted data.

    -It might make more sense to focus on radix sort, since that is not
already
    offered in the standard library...
Are you sure about the quicksort speed difference? That seems like a huge
difference. Did you have inlining enabled? Also, the current versions are
not the final forms. I am implementing random-access iterators and
user-provided functions.