[dagw]

I find this a recurring problem when trying to learn Haskell. On the one hand you have lots of books and tutorials talking about how simple and elegant Haskell is. Showing all the awesome things you can do with two lines of code. Then I try to write simple and elegant Haskell like that and find it runs an order of magnitude slower than my Python code.

I wish more Haskell proponents would spent less time showing off simple and elegant code, and more time showing fast and correct code.

[aristidb]

This is the real "sort" used by GHC 7: http://hackage.haskell.org/packages/archive/base/4.3.0.0/doc...

It's pretty elegant, too, if less than the inefficient pseudo-qsort shown by the OP.

[sizzla]

Correct me if I am wrong but they are actually using mergesort (mergeAll) as it is significantly faster in Haskell than in-place quicksort (qsort).

That means there is a lot of overhead for doing something simple like an in-place qsort that should be much faster than a mergesort.

I am at a loss on why the provided code is any more elegant than: http://en.literateprograms.org/Merge_sort_%28C_Plus_Plus%29

or this (also taken from rosetta code)?

  #include <iterator>
  #include <algorithm> // for std::partition
  #include <functional> // for std::less
  
  template<typename RandomAccessIterator,
           typename Order>
  void quicksort(RandomAccessIterator first, RandomAccessIterator last, Order order)
  {
    if (last - first > 1)
    {
      RandomAccessIterator split = std::partition(first+1, last, std::bind2nd(order, *first));
      std::iter_swap(first, split-1);
      quicksort(first, split-1, order);
      quicksort(split, last, order);
    }
  }

  template<typename RandomAccessIterator>
 void quicksort(RandomAccessIterator first, RandomAccessIterator last)
  {
    quicksort(first, last, std::less<typename std::iterator_traits<RandomAccessIterator>::value_type>());
  }

The above code is more verbose?

[merijnv]

> That means there is a lot of overhead for doing something simple like an in-place qsort

Well, the problem not so much that there is a lot of overhead for doing an in-place qsort. The problem is that "in-place" anything is impossible in Haskell, it violates referential transparency.

Merge sort however needs no in-place update and can trivially (in Haskell world) be parallelized (merge sort is associative, you can use it as the reduce step of MapReduce). So you are correct in concluding that mergesort is used in Haskell.

You state "something simple like an in-place qsort", your statement about qsort being simple makes some assumptions about your programming languages world, which are not the same for all languages especially not Haskell. No whether all this is a pro and a con depends on what sort of coding you're doing of course.

I think that people call Haskell's code more elegant because they (like me) find C++ template and type annotations to be annoyingly verbose and messy for no good reason. For example for the two quicksort functions:

"template<typename RandomAccessIterator> void quicksort(RandomAccessIterator first, RandomAccessIterator last)"

vs

"quicksort :: (Ord a) => [a] -> [a]" (where "Ord a" means that a is an instance of the typeclass of ordered items).

[sizzla]

If you want to be terse, you can just use typedefs but most of ISO C++03 libraries just assume your editor can autocomplete quickly. The culture is different and absolute clarity is preferred over terseness.

There's a lot of infrastructure behind MapReduce and I bet you any implementation is going to have a lot more code in it than what is in Data.List.

Once things are in RAM (say, less than 1 GB), quicksort is the fastest algorithm (Edit: or some other optimized in-place variation on the theme such as introsort). You can parallelize across cores all you want but all that is going to do is crap up your caches and cause bus contention, making your code more complex and slower. The bottleneck is your databus not the CPU, after all what is the cost of a compare in clock cycles? This is also why mergesort is slower and parallelizing is not going to fix it.

If you want to see things that are parallelizable check out vectorizing libraries such as Thrust that run on massively parallel GPGPUs that have the bus design and RAM to handle that kind of parallelization. They are written in C++ and both the implementation and the code to use them looks almost the same like the code I posted. The compiler technology of other languages is years behind and currently can't even target these kinds of architectures.

I've seen GPU massive parallelization in Haskell and all it did was to generate C code and then invoke the CUDA compiler. This was in a research paper. In the meantime you can fire up nvcc with Thrust and sort bazillion of keys with less than 10 lines of code in C++ using a well-tested library.

Edit: BTW, anyone (and I really don't know who would be doing this) who is even thinking of running a MapReduce Haskell cluster must have lots of free time and $$$ to burn on wasted CPU cycles.

[merijnv]

Well, I didn't mean an actual MapReduce cluster, that's just silly. I was just pointing out that Haskell compilers can convert trivially convert these sorts of things to multicore code. Parallel Haskell basically does this, converting normal code to easy multicore code.

My point wasn't about using Haskell for high performance computing, my point was that Haskell's compiler can convert normal Haskell code to multicore code without (much, if any) work from the programmer. There is a large gap in between sequential single threaded program and a full MapReduce cluster and I think there is a significant amount of software written in this gap where Haskell-like languages could help with the parallelism/concurrency. The only reason I referenced MapReduce is since its a well known example even for people who are not well versed in functional programming (its a very common concept in functional programming).

[aristidb]

No, because it is solving a different problem. Quicksort cannot sort linked-lists, not even in C++. I think most implementations of std::list<>::sort (in C++) use some variant of mergesort.

And it is not true that in-place modification is not possible in Haskell. It is just not considered elegant, but it can be encapsulated in a perfectly Haskell-y way. (Hint: ST monad.)

Also I did not claim that my linked code is more elegant than some other code. Where did I make that claim? So I don't have to defend the claim I did not make, right?